1548 lines
52 KiB
TypeScript
Raw Normal View History

import type { ClawdbotConfig, RuntimeEnv } from "openclaw/plugin-sdk/feishu";
import {
buildAgentMediaPayload,
buildPendingHistoryContextFromMap,
clearHistoryEntriesIfEnabled,
createScopedPairingAccess,
DEFAULT_GROUP_HISTORY_LIMIT,
type HistoryEntry,
issuePairingChallenge,
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
normalizeAgentId,
recordPendingHistoryEntryIfEnabled,
resolveOpenProviderRuntimeGroupPolicy,
resolveDefaultGroupPolicy,
warnMissingProviderGroupPolicyFallbackOnce,
} from "openclaw/plugin-sdk/feishu";
import { resolveFeishuAccount } from "./accounts.js";
import { createFeishuClient } from "./client.js";
import { tryRecordMessage, tryRecordMessagePersistent } from "./dedup.js";
import { maybeCreateDynamicAgent } from "./dynamic-agent.js";
import { normalizeFeishuExternalKey } from "./external-keys.js";
import { downloadMessageResourceFeishu } from "./media.js";
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
import { extractMentionTargets, isMentionForwardRequest } from "./mention.js";
import {
resolveFeishuGroupConfig,
resolveFeishuReplyPolicy,
resolveFeishuAllowlistMatch,
isFeishuGroupAllowed,
} from "./policy.js";
import { parsePostContent } from "./post.js";
import { createFeishuReplyDispatcher } from "./reply-dispatcher.js";
import { getFeishuRuntime } from "./runtime.js";
import { getMessageFeishu, sendMessageFeishu } from "./send.js";
import type { FeishuMessageContext, FeishuMediaInfo, ResolvedFeishuAccount } from "./types.js";
import type { DynamicAgentCreationConfig } from "./types.js";
// --- Permission error extraction ---
// Extract permission grant URL from Feishu API error response.
type PermissionError = {
code: number;
message: string;
grantUrl?: string;
};
const IGNORED_PERMISSION_SCOPE_TOKENS = ["contact:contact.base:readonly"];
// Feishu API sometimes returns incorrect scope names in permission error
// responses (e.g. "contact:contact.base:readonly" instead of the valid
// "contact:user.base:readonly"). This map corrects known mismatches.
const FEISHU_SCOPE_CORRECTIONS: Record<string, string> = {
"contact:contact.base:readonly": "contact:user.base:readonly",
};
function correctFeishuScopeInUrl(url: string): string {
let corrected = url;
for (const [wrong, right] of Object.entries(FEISHU_SCOPE_CORRECTIONS)) {
corrected = corrected.replaceAll(encodeURIComponent(wrong), encodeURIComponent(right));
corrected = corrected.replaceAll(wrong, right);
}
return corrected;
}
function shouldSuppressPermissionErrorNotice(permissionError: PermissionError): boolean {
const message = permissionError.message.toLowerCase();
return IGNORED_PERMISSION_SCOPE_TOKENS.some((token) => message.includes(token));
}
function extractPermissionError(err: unknown): PermissionError | null {
if (!err || typeof err !== "object") return null;
// Axios error structure: err.response.data contains the Feishu error
const axiosErr = err as { response?: { data?: unknown } };
const data = axiosErr.response?.data;
if (!data || typeof data !== "object") return null;
const feishuErr = data as {
code?: number;
msg?: string;
error?: { permission_violations?: Array<{ uri?: string }> };
};
// Feishu permission error code: 99991672
if (feishuErr.code !== 99991672) return null;
// Extract the grant URL from the error message (contains the direct link)
const msg = feishuErr.msg ?? "";
const urlMatch = msg.match(/https:\/\/[^\s,]+\/app\/[^\s,]+/);
const grantUrl = urlMatch?.[0] ? correctFeishuScopeInUrl(urlMatch[0]) : undefined;
return {
code: feishuErr.code,
message: msg,
grantUrl,
};
}
// --- Sender name resolution (so the agent can distinguish who is speaking in group chats) ---
// Cache display names by sender id (open_id/user_id) to avoid an API call on every message.
const SENDER_NAME_TTL_MS = 10 * 60 * 1000;
const senderNameCache = new Map<string, { name: string; expireAt: number }>();
// Cache permission errors to avoid spamming the user with repeated notifications.
// Key: appId or "default", Value: timestamp of last notification
const permissionErrorNotifiedAt = new Map<string, number>();
const PERMISSION_ERROR_COOLDOWN_MS = 5 * 60 * 1000; // 5 minutes
type SenderNameResult = {
name?: string;
permissionError?: PermissionError;
};
function resolveSenderLookupIdType(senderId: string): "open_id" | "user_id" | "union_id" {
const trimmed = senderId.trim();
if (trimmed.startsWith("ou_")) {
return "open_id";
}
if (trimmed.startsWith("on_")) {
return "union_id";
}
return "user_id";
}
async function resolveFeishuSenderName(params: {
account: ResolvedFeishuAccount;
senderId: string;
log: (...args: any[]) => void;
}): Promise<SenderNameResult> {
const { account, senderId, log } = params;
if (!account.configured) return {};
const normalizedSenderId = senderId.trim();
if (!normalizedSenderId) return {};
const cached = senderNameCache.get(normalizedSenderId);
const now = Date.now();
if (cached && cached.expireAt > now) return { name: cached.name };
try {
const client = createFeishuClient(account);
const userIdType = resolveSenderLookupIdType(normalizedSenderId);
// contact/v3/users/:user_id?user_id_type=<open_id|user_id|union_id>
const res: any = await client.contact.user.get({
path: { user_id: normalizedSenderId },
params: { user_id_type: userIdType },
});
const name: string | undefined =
res?.data?.user?.name ||
res?.data?.user?.display_name ||
res?.data?.user?.nickname ||
res?.data?.user?.en_name;
if (name && typeof name === "string") {
senderNameCache.set(normalizedSenderId, { name, expireAt: now + SENDER_NAME_TTL_MS });
return { name };
}
return {};
} catch (err) {
// Check if this is a permission error
const permErr = extractPermissionError(err);
if (permErr) {
if (shouldSuppressPermissionErrorNotice(permErr)) {
log(`feishu: ignoring stale permission scope error: ${permErr.message}`);
return {};
}
log(`feishu: permission error resolving sender name: code=${permErr.code}`);
return { permissionError: permErr };
}
// Best-effort. Don't fail message handling if name lookup fails.
log(`feishu: failed to resolve sender name for ${normalizedSenderId}: ${String(err)}`);
return {};
}
}
export type FeishuMessageEvent = {
sender: {
sender_id: {
open_id?: string;
user_id?: string;
union_id?: string;
};
sender_type?: string;
tenant_key?: string;
};
message: {
message_id: string;
root_id?: string;
parent_id?: string;
thread_id?: string;
chat_id: string;
chat_type: "p2p" | "group" | "private";
message_type: string;
content: string;
create_time?: string;
mentions?: Array<{
key: string;
id: {
open_id?: string;
user_id?: string;
union_id?: string;
};
name: string;
tenant_key?: string;
}>;
};
};
export type FeishuBotAddedEvent = {
chat_id: string;
operator_id: {
open_id?: string;
user_id?: string;
union_id?: string;
};
external: boolean;
operator_tenant_key?: string;
};
type GroupSessionScope = "group" | "group_sender" | "group_topic" | "group_topic_sender";
type ResolvedFeishuGroupSession = {
peerId: string;
parentPeer: { kind: "group"; id: string } | null;
groupSessionScope: GroupSessionScope;
replyInThread: boolean;
threadReply: boolean;
};
function resolveFeishuGroupSession(params: {
chatId: string;
senderOpenId: string;
messageId: string;
rootId?: string;
threadId?: string;
groupConfig?: {
groupSessionScope?: GroupSessionScope;
topicSessionMode?: "enabled" | "disabled";
replyInThread?: "enabled" | "disabled";
};
feishuCfg?: {
groupSessionScope?: GroupSessionScope;
topicSessionMode?: "enabled" | "disabled";
replyInThread?: "enabled" | "disabled";
};
}): ResolvedFeishuGroupSession {
const { chatId, senderOpenId, messageId, rootId, threadId, groupConfig, feishuCfg } = params;
const normalizedThreadId = threadId?.trim();
const normalizedRootId = rootId?.trim();
const threadReply = Boolean(normalizedThreadId || normalizedRootId);
const replyInThread =
(groupConfig?.replyInThread ?? feishuCfg?.replyInThread ?? "disabled") === "enabled" ||
threadReply;
const legacyTopicSessionMode =
groupConfig?.topicSessionMode ?? feishuCfg?.topicSessionMode ?? "disabled";
const groupSessionScope: GroupSessionScope =
groupConfig?.groupSessionScope ??
feishuCfg?.groupSessionScope ??
(legacyTopicSessionMode === "enabled" ? "group_topic" : "group");
// Keep topic session keys stable across the "first turn creates thread" flow:
// first turn may only have message_id, while the next turn carries root_id/thread_id.
// Prefer root_id first so both turns stay on the same peer key.
const topicScope =
groupSessionScope === "group_topic" || groupSessionScope === "group_topic_sender"
? (normalizedRootId ?? normalizedThreadId ?? (replyInThread ? messageId : null))
: null;
let peerId = chatId;
switch (groupSessionScope) {
case "group_sender":
peerId = `${chatId}:sender:${senderOpenId}`;
break;
case "group_topic":
peerId = topicScope ? `${chatId}:topic:${topicScope}` : chatId;
break;
case "group_topic_sender":
peerId = topicScope
? `${chatId}:topic:${topicScope}:sender:${senderOpenId}`
: `${chatId}:sender:${senderOpenId}`;
break;
case "group":
default:
peerId = chatId;
break;
}
const parentPeer =
topicScope &&
(groupSessionScope === "group_topic" || groupSessionScope === "group_topic_sender")
? {
kind: "group" as const,
id: chatId,
}
: null;
return {
peerId,
parentPeer,
groupSessionScope,
replyInThread,
threadReply,
};
}
function parseMessageContent(content: string, messageType: string): string {
if (messageType === "post") {
// Extract text content from rich text post
const { textContent } = parsePostContent(content);
return textContent;
}
try {
const parsed = JSON.parse(content);
if (messageType === "text") {
return parsed.text || "";
}
if (messageType === "share_chat") {
// Preserve available summary text for merged/forwarded chat messages.
if (parsed && typeof parsed === "object") {
const share = parsed as {
body?: unknown;
summary?: unknown;
share_chat_id?: unknown;
};
if (typeof share.body === "string" && share.body.trim().length > 0) {
return share.body.trim();
}
if (typeof share.summary === "string" && share.summary.trim().length > 0) {
return share.summary.trim();
}
if (typeof share.share_chat_id === "string" && share.share_chat_id.trim().length > 0) {
return `[Forwarded message: ${share.share_chat_id.trim()}]`;
}
}
return "[Forwarded message]";
}
if (messageType === "merge_forward") {
// Return placeholder; actual content fetched asynchronously in handleFeishuMessage
return "[Merged and Forwarded Message - loading...]";
}
return content;
} catch {
return content;
}
}
/**
* Parse merge_forward message content and fetch sub-messages.
* Returns formatted text content of all sub-messages.
*/
function parseMergeForwardContent(params: {
content: string;
log?: (...args: any[]) => void;
}): string {
const { content, log } = params;
const maxMessages = 50;
// For merge_forward, the API returns all sub-messages in items array
// with upper_message_id pointing to the merge_forward message.
// The 'content' parameter here is actually the full API response items array as JSON.
log?.(`feishu: parsing merge_forward sub-messages from API response`);
let items: Array<{
message_id?: string;
msg_type?: string;
body?: { content?: string };
sender?: { id?: string };
upper_message_id?: string;
create_time?: string;
}>;
try {
items = JSON.parse(content);
} catch {
log?.(`feishu: merge_forward items parse failed`);
return "[Merged and Forwarded Message - parse error]";
}
if (!Array.isArray(items) || items.length === 0) {
return "[Merged and Forwarded Message - no sub-messages]";
}
// Filter to only sub-messages (those with upper_message_id, skip the merge_forward container itself)
const subMessages = items.filter((item) => item.upper_message_id);
if (subMessages.length === 0) {
return "[Merged and Forwarded Message - no sub-messages found]";
}
log?.(`feishu: merge_forward contains ${subMessages.length} sub-messages`);
// Sort by create_time
subMessages.sort((a, b) => {
const timeA = parseInt(a.create_time || "0", 10);
const timeB = parseInt(b.create_time || "0", 10);
return timeA - timeB;
});
// Format output
const lines: string[] = ["[Merged and Forwarded Messages]"];
const limitedMessages = subMessages.slice(0, maxMessages);
for (const item of limitedMessages) {
const msgContent = item.body?.content || "";
const msgType = item.msg_type || "text";
const formatted = formatSubMessageContent(msgContent, msgType);
lines.push(`- ${formatted}`);
}
if (subMessages.length > maxMessages) {
lines.push(`... and ${subMessages.length - maxMessages} more messages`);
}
return lines.join("\n");
}
/**
* Format sub-message content based on message type.
*/
function formatSubMessageContent(content: string, contentType: string): string {
try {
const parsed = JSON.parse(content);
switch (contentType) {
case "text":
return parsed.text || content;
case "post": {
const { textContent } = parsePostContent(content);
return textContent;
}
case "image":
return "[Image]";
case "file":
return `[File: ${parsed.file_name || "unknown"}]`;
case "audio":
return "[Audio]";
case "video":
return "[Video]";
case "sticker":
return "[Sticker]";
case "merge_forward":
return "[Nested Merged Forward]";
default:
return `[${contentType}]`;
}
} catch {
return content;
}
}
function checkBotMentioned(event: FeishuMessageEvent, botOpenId?: string): boolean {
if (!botOpenId) return false;
fix(feishu): guard against false-positive @mentions in multi-app groups (#30315) * fix(feishu): guard against false-positive @mentions in multi-app groups When multiple Feishu bot apps share a group chat, Feishu's WebSocket event delivery remaps the open_id in mentions[] per-app. This causes checkBotMentioned() to return true for ALL bots when only one was actually @mentioned, making requireMention ineffective. Add a botName guard: if the mention's open_id matches this bot but the mention's display name differs from this bot's configured botName, treat it as a false positive and skip. botName is already available via account.config.botName (set during onboarding). Closes #24249 * fix(feishu): support @all mention in multi-bot groups When a user sends @all (@_all in Feishu message content), treat it as mentioning every bot so all agents respond when requireMention is true. Feishu's @all does not populate the mentions[] array, so this needs explicit content-level detection. * fix(feishu): auto-fetch bot display name from API for reliable mention matching Instead of relying on the manually configured botName (which may differ from the actual Feishu bot display name), fetch the bot's display name from the Feishu API at startup via probeFeishu(). This ensures checkBotMentioned() always compares against the correct display name, even when the config botName doesn't match (e.g. config says 'Wanda' but Feishu shows '绯红女巫'). Changes: - monitor.ts: fetchBotOpenId → fetchBotInfo (returns both openId and name) - monitor.ts: store botNames map, pass botName to handleFeishuMessage - bot.ts: accept botName from params, prefer it over config fallback * Changelog: note Feishu multi-app mention false-positive guard --------- Co-authored-by: Teague Xiao <teaguexiao@TeaguedeMac-mini.local> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:27:35 +08:00
// Check for @all (@_all in Feishu) — treat as mentioning every bot
const rawContent = event.message.content ?? "";
if (rawContent.includes("@_all")) return true;
const mentions = event.message.mentions ?? [];
if (mentions.length > 0) {
// Rely on Feishu mention IDs; display names can vary by alias/context.
return mentions.some((m) => m.id.open_id === botOpenId);
}
// Post (rich text) messages may have empty message.mentions when they contain docs/paste
if (event.message.message_type === "post") {
2026-02-16 21:41:57 +08:00
const { mentionedOpenIds } = parsePostContent(event.message.content);
return mentionedOpenIds.some((id) => id === botOpenId);
}
return false;
}
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
function normalizeMentions(
text: string,
mentions?: FeishuMessageEvent["message"]["mentions"],
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
botStripId?: string,
): string {
if (!mentions || mentions.length === 0) return text;
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
const escaped = (value: string) => value.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
const escapeName = (value: string) => value.replace(/</g, "&lt;").replace(/>/g, "&gt;");
let result = text;
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
for (const mention of mentions) {
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
const mentionId = mention.id.open_id;
const replacement =
botStripId && mentionId === botStripId
? ""
: mentionId
? `<at user_id="${mentionId}">${escapeName(mention.name)}</at>`
: `@${mention.name}`;
result = result.replace(new RegExp(escaped(mention.key), "g"), () => replacement).trim();
}
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
return result;
}
function normalizeFeishuCommandProbeBody(text: string): string {
if (!text) {
return "";
}
return text
.replace(/<at\b[^>]*>[^<]*<\/at>/giu, " ")
.replace(/(^|\s)@[^/\s]+(?=\s|$|\/)/gu, "$1")
.replace(/\s+/g, " ")
.trim();
}
/**
* Parse media keys from message content based on message type.
*/
function parseMediaKeys(
content: string,
messageType: string,
): {
imageKey?: string;
fileKey?: string;
fileName?: string;
} {
try {
const parsed = JSON.parse(content);
const imageKey = normalizeFeishuExternalKey(parsed.image_key);
const fileKey = normalizeFeishuExternalKey(parsed.file_key);
switch (messageType) {
case "image":
return { imageKey };
case "file":
return { fileKey, fileName: parsed.file_name };
case "audio":
return { fileKey };
case "video":
case "media":
// Video/media has both file_key (video) and image_key (thumbnail)
return { fileKey, imageKey };
case "sticker":
return { fileKey };
default:
return {};
}
} catch {
return {};
}
}
/**
* Map Feishu message type to messageResource.get resource type.
* Feishu messageResource API supports only: image | file.
*/
export function toMessageResourceType(messageType: string): "image" | "file" {
return messageType === "image" ? "image" : "file";
}
/**
* Infer placeholder text based on message type.
*/
function inferPlaceholder(messageType: string): string {
switch (messageType) {
case "image":
return "<media:image>";
case "file":
return "<media:document>";
case "audio":
return "<media:audio>";
case "video":
case "media":
return "<media:video>";
case "sticker":
return "<media:sticker>";
default:
return "<media:document>";
}
}
/**
* Resolve media from a Feishu message, downloading and saving to disk.
* Similar to Discord's resolveMediaList().
*/
async function resolveFeishuMediaList(params: {
cfg: ClawdbotConfig;
messageId: string;
messageType: string;
content: string;
maxBytes: number;
log?: (msg: string) => void;
accountId?: string;
}): Promise<FeishuMediaInfo[]> {
const { cfg, messageId, messageType, content, maxBytes, log, accountId } = params;
// Only process media message types (including post for embedded images)
const mediaTypes = ["image", "file", "audio", "video", "media", "sticker", "post"];
if (!mediaTypes.includes(messageType)) {
return [];
}
const out: FeishuMediaInfo[] = [];
const core = getFeishuRuntime();
// Handle post (rich text) messages with embedded images/media.
if (messageType === "post") {
const { imageKeys, mediaKeys: postMediaKeys } = parsePostContent(content);
if (imageKeys.length === 0 && postMediaKeys.length === 0) {
return [];
}
if (imageKeys.length > 0) {
log?.(`feishu: post message contains ${imageKeys.length} embedded image(s)`);
}
if (postMediaKeys.length > 0) {
log?.(`feishu: post message contains ${postMediaKeys.length} embedded media file(s)`);
}
for (const imageKey of imageKeys) {
try {
// Embedded images in post use messageResource API with image_key as file_key
const result = await downloadMessageResourceFeishu({
cfg,
messageId,
fileKey: imageKey,
type: "image",
accountId,
});
let contentType = result.contentType;
if (!contentType) {
contentType = await core.media.detectMime({ buffer: result.buffer });
}
const saved = await core.channel.media.saveMediaBuffer(
result.buffer,
contentType,
"inbound",
maxBytes,
);
out.push({
path: saved.path,
contentType: saved.contentType,
placeholder: "<media:image>",
});
log?.(`feishu: downloaded embedded image ${imageKey}, saved to ${saved.path}`);
} catch (err) {
log?.(`feishu: failed to download embedded image ${imageKey}: ${String(err)}`);
}
}
for (const media of postMediaKeys) {
try {
const result = await downloadMessageResourceFeishu({
cfg,
messageId,
fileKey: media.fileKey,
type: "file",
accountId,
});
let contentType = result.contentType;
if (!contentType) {
contentType = await core.media.detectMime({ buffer: result.buffer });
}
const saved = await core.channel.media.saveMediaBuffer(
result.buffer,
contentType,
"inbound",
maxBytes,
);
out.push({
path: saved.path,
contentType: saved.contentType,
placeholder: "<media:video>",
});
log?.(`feishu: downloaded embedded media ${media.fileKey}, saved to ${saved.path}`);
} catch (err) {
log?.(`feishu: failed to download embedded media ${media.fileKey}: ${String(err)}`);
}
}
return out;
}
// Handle other media types
const mediaKeys = parseMediaKeys(content, messageType);
if (!mediaKeys.imageKey && !mediaKeys.fileKey) {
return [];
}
try {
let buffer: Buffer;
let contentType: string | undefined;
let fileName: string | undefined;
// For message media, always use messageResource API
// The image.get API is only for images uploaded via im/v1/images, not for message attachments
const fileKey = mediaKeys.fileKey || mediaKeys.imageKey;
if (!fileKey) {
return [];
}
const resourceType = toMessageResourceType(messageType);
const result = await downloadMessageResourceFeishu({
cfg,
messageId,
fileKey,
type: resourceType,
accountId,
});
buffer = result.buffer;
contentType = result.contentType;
fileName = result.fileName || mediaKeys.fileName;
// Detect mime type if not provided
if (!contentType) {
contentType = await core.media.detectMime({ buffer });
}
// Save to disk using core's saveMediaBuffer
const saved = await core.channel.media.saveMediaBuffer(
buffer,
contentType,
"inbound",
maxBytes,
fileName,
);
out.push({
path: saved.path,
contentType: saved.contentType,
placeholder: inferPlaceholder(messageType),
});
log?.(`feishu: downloaded ${messageType} media, saved to ${saved.path}`);
} catch (err) {
log?.(`feishu: failed to download ${messageType} media: ${String(err)}`);
}
return out;
}
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
// --- Broadcast support ---
// Resolve broadcast agent list for a given peer (group) ID.
// Returns null if no broadcast config exists or the peer is not in the broadcast list.
export function resolveBroadcastAgents(cfg: ClawdbotConfig, peerId: string): string[] | null {
const broadcast = (cfg as Record<string, unknown>).broadcast;
if (!broadcast || typeof broadcast !== "object") return null;
const agents = (broadcast as Record<string, unknown>)[peerId];
if (!Array.isArray(agents) || agents.length === 0) return null;
return agents as string[];
}
// Build a session key for a broadcast target agent by replacing the agent ID prefix.
// Session keys follow the format: agent:<agentId>:<channel>:<peerKind>:<peerId>
export function buildBroadcastSessionKey(
baseSessionKey: string,
originalAgentId: string,
targetAgentId: string,
): string {
const prefix = `agent:${originalAgentId}:`;
if (baseSessionKey.startsWith(prefix)) {
return `agent:${targetAgentId}:${baseSessionKey.slice(prefix.length)}`;
}
return baseSessionKey;
}
/**
* Build media payload for inbound context.
* Similar to Discord's buildDiscordMediaPayload().
*/
export function parseFeishuMessageEvent(
event: FeishuMessageEvent,
botOpenId?: string,
_botName?: string,
): FeishuMessageContext {
const rawContent = parseMessageContent(event.message.content, event.message.message_type);
const mentionedBot = checkBotMentioned(event, botOpenId);
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
const hasAnyMention = (event.message.mentions?.length ?? 0) > 0;
// Strip the bot's own mention so slash commands like @Bot /help retain
// the leading /. This applies in both p2p *and* group contexts — the
// mentionedBot flag already captures whether the bot was addressed, so
// keeping the mention tag in content only breaks command detection (#35994).
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
// Non-bot mentions (e.g. mention-forward targets) are still normalized to <at> tags.
const content = normalizeMentions(rawContent, event.message.mentions, botOpenId);
const senderOpenId = event.sender.sender_id.open_id?.trim();
const senderUserId = event.sender.sender_id.user_id?.trim();
const senderFallbackId = senderOpenId || senderUserId || "";
const ctx: FeishuMessageContext = {
chatId: event.message.chat_id,
messageId: event.message.message_id,
senderId: senderUserId || senderOpenId || "",
// Keep the historical field name, but fall back to user_id when open_id is unavailable
// (common in some mobile app deliveries).
senderOpenId: senderFallbackId,
chatType: event.message.chat_type,
mentionedBot,
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
hasAnyMention,
rootId: event.message.root_id || undefined,
parentId: event.message.parent_id || undefined,
threadId: event.message.thread_id || undefined,
content,
contentType: event.message.message_type,
};
// Detect mention forward request: message mentions bot + at least one other user
if (isMentionForwardRequest(event, botOpenId)) {
const mentionTargets = extractMentionTargets(event, botOpenId);
if (mentionTargets.length > 0) {
ctx.mentionTargets = mentionTargets;
}
}
return ctx;
}
export function buildFeishuAgentBody(params: {
ctx: Pick<
FeishuMessageContext,
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
"content" | "senderName" | "senderOpenId" | "mentionTargets" | "messageId" | "hasAnyMention"
>;
quotedContent?: string;
permissionErrorForAgent?: PermissionError;
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
botOpenId?: string;
}): string {
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
const { ctx, quotedContent, permissionErrorForAgent, botOpenId } = params;
let messageBody = ctx.content;
if (quotedContent) {
messageBody = `[Replying to: "${quotedContent}"]\n\n${ctx.content}`;
}
// DMs already have per-sender sessions, but this label still improves attribution.
const speaker = ctx.senderName ?? ctx.senderOpenId;
messageBody = `${speaker}: ${messageBody}`;
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
if (ctx.hasAnyMention) {
const botIdHint = botOpenId?.trim();
messageBody +=
`\n\n[System: The content may include mention tags in the form <at user_id="...">name</at>. ` +
`Treat these as real mentions of Feishu entities (users or bots).]`;
if (botIdHint) {
messageBody += `\n[System: If user_id is "${botIdHint}", that mention refers to you.]`;
}
}
if (ctx.mentionTargets && ctx.mentionTargets.length > 0) {
const targetNames = ctx.mentionTargets.map((t) => t.name).join(", ");
messageBody += `\n\n[System: Your reply will automatically @mention: ${targetNames}. Do not write @xxx yourself.]`;
}
// Keep message_id on its own line so shared message-id hint stripping can parse it reliably.
messageBody = `[message_id: ${ctx.messageId}]\n${messageBody}`;
if (permissionErrorForAgent) {
const grantUrl = permissionErrorForAgent.grantUrl ?? "";
messageBody += `\n\n[System: The bot encountered a Feishu API permission error. Please inform the user about this issue and provide the permission grant URL for the admin to authorize. Permission grant URL: ${grantUrl}]`;
}
return messageBody;
}
export async function handleFeishuMessage(params: {
cfg: ClawdbotConfig;
event: FeishuMessageEvent;
botOpenId?: string;
fix(feishu): guard against false-positive @mentions in multi-app groups (#30315) * fix(feishu): guard against false-positive @mentions in multi-app groups When multiple Feishu bot apps share a group chat, Feishu's WebSocket event delivery remaps the open_id in mentions[] per-app. This causes checkBotMentioned() to return true for ALL bots when only one was actually @mentioned, making requireMention ineffective. Add a botName guard: if the mention's open_id matches this bot but the mention's display name differs from this bot's configured botName, treat it as a false positive and skip. botName is already available via account.config.botName (set during onboarding). Closes #24249 * fix(feishu): support @all mention in multi-bot groups When a user sends @all (@_all in Feishu message content), treat it as mentioning every bot so all agents respond when requireMention is true. Feishu's @all does not populate the mentions[] array, so this needs explicit content-level detection. * fix(feishu): auto-fetch bot display name from API for reliable mention matching Instead of relying on the manually configured botName (which may differ from the actual Feishu bot display name), fetch the bot's display name from the Feishu API at startup via probeFeishu(). This ensures checkBotMentioned() always compares against the correct display name, even when the config botName doesn't match (e.g. config says 'Wanda' but Feishu shows '绯红女巫'). Changes: - monitor.ts: fetchBotOpenId → fetchBotInfo (returns both openId and name) - monitor.ts: store botNames map, pass botName to handleFeishuMessage - bot.ts: accept botName from params, prefer it over config fallback * Changelog: note Feishu multi-app mention false-positive guard --------- Co-authored-by: Teague Xiao <teaguexiao@TeaguedeMac-mini.local> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:27:35 +08:00
botName?: string;
runtime?: RuntimeEnv;
chatHistories?: Map<string, HistoryEntry[]>;
accountId?: string;
}): Promise<void> {
fix(feishu): guard against false-positive @mentions in multi-app groups (#30315) * fix(feishu): guard against false-positive @mentions in multi-app groups When multiple Feishu bot apps share a group chat, Feishu's WebSocket event delivery remaps the open_id in mentions[] per-app. This causes checkBotMentioned() to return true for ALL bots when only one was actually @mentioned, making requireMention ineffective. Add a botName guard: if the mention's open_id matches this bot but the mention's display name differs from this bot's configured botName, treat it as a false positive and skip. botName is already available via account.config.botName (set during onboarding). Closes #24249 * fix(feishu): support @all mention in multi-bot groups When a user sends @all (@_all in Feishu message content), treat it as mentioning every bot so all agents respond when requireMention is true. Feishu's @all does not populate the mentions[] array, so this needs explicit content-level detection. * fix(feishu): auto-fetch bot display name from API for reliable mention matching Instead of relying on the manually configured botName (which may differ from the actual Feishu bot display name), fetch the bot's display name from the Feishu API at startup via probeFeishu(). This ensures checkBotMentioned() always compares against the correct display name, even when the config botName doesn't match (e.g. config says 'Wanda' but Feishu shows '绯红女巫'). Changes: - monitor.ts: fetchBotOpenId → fetchBotInfo (returns both openId and name) - monitor.ts: store botNames map, pass botName to handleFeishuMessage - bot.ts: accept botName from params, prefer it over config fallback * Changelog: note Feishu multi-app mention false-positive guard --------- Co-authored-by: Teague Xiao <teaguexiao@TeaguedeMac-mini.local> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:27:35 +08:00
const { cfg, event, botOpenId, botName, runtime, chatHistories, accountId } = params;
// Resolve account with merged config
const account = resolveFeishuAccount({ cfg, accountId });
const feishuCfg = account.config;
const log = runtime?.log ?? console.log;
const error = runtime?.error ?? console.error;
// Dedup: synchronous memory guard prevents concurrent duplicate dispatch
// before the async persistent check completes.
const messageId = event.message.message_id;
const memoryDedupeKey = `${account.accountId}:${messageId}`;
if (!tryRecordMessage(memoryDedupeKey)) {
log(`feishu: skipping duplicate message ${messageId} (memory dedup)`);
return;
}
// Persistent dedup survives restarts and reconnects.
if (!(await tryRecordMessagePersistent(messageId, account.accountId, log))) {
log(`feishu: skipping duplicate message ${messageId}`);
return;
}
let ctx = parseFeishuMessageEvent(event, botOpenId, botName);
const isGroup = ctx.chatType === "group";
const isDirect = !isGroup;
const senderUserId = event.sender.sender_id.user_id?.trim() || undefined;
// Handle merge_forward messages: fetch full message via API then expand sub-messages
if (event.message.message_type === "merge_forward") {
log(
`feishu[${account.accountId}]: processing merge_forward message, fetching full content via API`,
);
try {
// Websocket event doesn't include sub-messages, need to fetch via API
// The API returns all sub-messages in the items array
const client = createFeishuClient(account);
const response = (await client.im.message.get({
path: { message_id: event.message.message_id },
})) as { code?: number; data?: { items?: unknown[] } };
if (response.code === 0 && response.data?.items && response.data.items.length > 0) {
log(
`feishu[${account.accountId}]: merge_forward API returned ${response.data.items.length} items`,
);
const expandedContent = parseMergeForwardContent({
content: JSON.stringify(response.data.items),
log,
});
ctx = { ...ctx, content: expandedContent };
} else {
log(`feishu[${account.accountId}]: merge_forward API returned no items`);
ctx = { ...ctx, content: "[Merged and Forwarded Message - could not fetch]" };
}
} catch (err) {
log(`feishu[${account.accountId}]: merge_forward fetch failed: ${String(err)}`);
ctx = { ...ctx, content: "[Merged and Forwarded Message - fetch error]" };
}
}
// Resolve sender display name (best-effort) so the agent can attribute messages correctly.
// Optimization: skip if disabled to save API quota (Feishu free tier limit).
let permissionErrorForAgent: PermissionError | undefined;
if (feishuCfg?.resolveSenderNames ?? true) {
const senderResult = await resolveFeishuSenderName({
account,
senderId: ctx.senderOpenId,
log,
});
if (senderResult.name) ctx = { ...ctx, senderName: senderResult.name };
// Track permission error to inform agent later (with cooldown to avoid repetition)
if (senderResult.permissionError) {
const appKey = account.appId ?? "default";
const now = Date.now();
const lastNotified = permissionErrorNotifiedAt.get(appKey) ?? 0;
if (now - lastNotified > PERMISSION_ERROR_COOLDOWN_MS) {
permissionErrorNotifiedAt.set(appKey, now);
permissionErrorForAgent = senderResult.permissionError;
}
}
}
log(
`feishu[${account.accountId}]: received message from ${ctx.senderOpenId} in ${ctx.chatId} (${ctx.chatType})`,
);
// Log mention targets if detected
if (ctx.mentionTargets && ctx.mentionTargets.length > 0) {
const names = ctx.mentionTargets.map((t) => t.name).join(", ");
log(`feishu[${account.accountId}]: detected @ forward request, targets: [${names}]`);
}
const historyLimit = Math.max(
0,
feishuCfg?.historyLimit ?? cfg.messages?.groupChat?.historyLimit ?? DEFAULT_GROUP_HISTORY_LIMIT,
);
const groupConfig = isGroup
? resolveFeishuGroupConfig({ cfg: feishuCfg, groupId: ctx.chatId })
: undefined;
const groupSession = isGroup
? resolveFeishuGroupSession({
chatId: ctx.chatId,
senderOpenId: ctx.senderOpenId,
messageId: ctx.messageId,
rootId: ctx.rootId,
threadId: ctx.threadId,
groupConfig,
feishuCfg,
})
: null;
const groupHistoryKey = isGroup ? (groupSession?.peerId ?? ctx.chatId) : undefined;
const dmPolicy = feishuCfg?.dmPolicy ?? "pairing";
const configAllowFrom = feishuCfg?.allowFrom ?? [];
const useAccessGroups = cfg.commands?.useAccessGroups !== false;
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
const rawBroadcastAgents = isGroup ? resolveBroadcastAgents(cfg, ctx.chatId) : null;
const broadcastAgents = rawBroadcastAgents
? [...new Set(rawBroadcastAgents.map((id) => normalizeAgentId(id)))]
: null;
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
let requireMention = false; // DMs never require mention; groups may override below
if (isGroup) {
if (groupConfig?.enabled === false) {
log(`feishu[${account.accountId}]: group ${ctx.chatId} is disabled`);
return;
}
const defaultGroupPolicy = resolveDefaultGroupPolicy(cfg);
const { groupPolicy, providerMissingFallbackApplied } = resolveOpenProviderRuntimeGroupPolicy({
providerConfigPresent: cfg.channels?.feishu !== undefined,
groupPolicy: feishuCfg?.groupPolicy,
defaultGroupPolicy,
});
warnMissingProviderGroupPolicyFallbackOnce({
providerMissingFallbackApplied,
providerKey: "feishu",
accountId: account.accountId,
log,
});
const groupAllowFrom = feishuCfg?.groupAllowFrom ?? [];
// DEBUG: log(`feishu[${account.accountId}]: groupPolicy=${groupPolicy}`);
// Check if this GROUP is allowed (groupAllowFrom contains group IDs like oc_xxx, not user IDs)
const groupAllowed = isFeishuGroupAllowed({
groupPolicy,
allowFrom: groupAllowFrom,
senderId: ctx.chatId, // Check group ID, not sender ID
senderName: undefined,
});
if (!groupAllowed) {
log(
`feishu[${account.accountId}]: group ${ctx.chatId} not in groupAllowFrom (groupPolicy=${groupPolicy})`,
);
return;
}
// Sender-level allowlist: per-group allowFrom takes precedence, then global groupSenderAllowFrom
const perGroupSenderAllowFrom = groupConfig?.allowFrom ?? [];
const globalSenderAllowFrom = feishuCfg?.groupSenderAllowFrom ?? [];
const effectiveSenderAllowFrom =
perGroupSenderAllowFrom.length > 0 ? perGroupSenderAllowFrom : globalSenderAllowFrom;
if (effectiveSenderAllowFrom.length > 0) {
const senderAllowed = isFeishuGroupAllowed({
groupPolicy: "allowlist",
allowFrom: effectiveSenderAllowFrom,
senderId: ctx.senderOpenId,
senderIds: [senderUserId],
senderName: ctx.senderName,
});
if (!senderAllowed) {
log(`feishu: sender ${ctx.senderOpenId} not in group ${ctx.chatId} sender allowlist`);
return;
}
}
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
({ requireMention } = resolveFeishuReplyPolicy({
isDirectMessage: false,
globalConfig: feishuCfg,
groupConfig,
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
}));
if (requireMention && !ctx.mentionedBot) {
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
log(`feishu[${account.accountId}]: message in group ${ctx.chatId} did not mention bot`);
// Record to pending history for non-broadcast groups only. For broadcast groups,
// the mentioned handler's broadcast dispatch writes the turn directly into all
// agent sessions — buffering here would cause duplicate replay when this account
// later becomes active via buildPendingHistoryContextFromMap.
if (!broadcastAgents && chatHistories && groupHistoryKey) {
recordPendingHistoryEntryIfEnabled({
historyMap: chatHistories,
historyKey: groupHistoryKey,
limit: historyLimit,
entry: {
sender: ctx.senderOpenId,
body: `${ctx.senderName ?? ctx.senderOpenId}: ${ctx.content}`,
timestamp: Date.now(),
messageId: ctx.messageId,
},
});
}
return;
}
} else {
}
try {
const core = getFeishuRuntime();
const pairing = createScopedPairingAccess({
core,
channel: "feishu",
accountId: account.accountId,
});
const commandProbeBody = isGroup ? normalizeFeishuCommandProbeBody(ctx.content) : ctx.content;
const shouldComputeCommandAuthorized = core.channel.commands.shouldComputeCommandAuthorized(
commandProbeBody,
cfg,
);
const storeAllowFrom =
!isGroup &&
dmPolicy !== "allowlist" &&
(dmPolicy !== "open" || shouldComputeCommandAuthorized)
? await pairing.readAllowFromStore().catch(() => [])
: [];
const effectiveDmAllowFrom = [...configAllowFrom, ...storeAllowFrom];
const dmAllowed = resolveFeishuAllowlistMatch({
allowFrom: effectiveDmAllowFrom,
senderId: ctx.senderOpenId,
senderIds: [senderUserId],
senderName: ctx.senderName,
}).allowed;
if (isDirect && dmPolicy !== "open" && !dmAllowed) {
if (dmPolicy === "pairing") {
await issuePairingChallenge({
channel: "feishu",
senderId: ctx.senderOpenId,
senderIdLine: `Your Feishu user id: ${ctx.senderOpenId}`,
meta: { name: ctx.senderName },
upsertPairingRequest: pairing.upsertPairingRequest,
onCreated: () => {
log(`feishu[${account.accountId}]: pairing request sender=${ctx.senderOpenId}`);
},
sendPairingReply: async (text) => {
await sendMessageFeishu({
cfg,
to: `chat:${ctx.chatId}`,
text,
accountId: account.accountId,
});
},
onReplyError: (err) => {
log(
`feishu[${account.accountId}]: pairing reply failed for ${ctx.senderOpenId}: ${String(err)}`,
);
},
});
} else {
log(
`feishu[${account.accountId}]: blocked unauthorized sender ${ctx.senderOpenId} (dmPolicy=${dmPolicy})`,
);
}
return;
}
const commandAllowFrom = isGroup
? (groupConfig?.allowFrom ?? configAllowFrom)
: effectiveDmAllowFrom;
const senderAllowedForCommands = resolveFeishuAllowlistMatch({
allowFrom: commandAllowFrom,
senderId: ctx.senderOpenId,
senderIds: [senderUserId],
senderName: ctx.senderName,
}).allowed;
const commandAuthorized = shouldComputeCommandAuthorized
? core.channel.commands.resolveCommandAuthorizedFromAuthorizers({
useAccessGroups,
authorizers: [
{ configured: commandAllowFrom.length > 0, allowed: senderAllowedForCommands },
],
})
: undefined;
// In group chats, the session is scoped to the group, but the *speaker* is the sender.
// Using a group-scoped From causes the agent to treat different users as the same person.
const feishuFrom = `feishu:${ctx.senderOpenId}`;
const feishuTo = isGroup ? `chat:${ctx.chatId}` : `user:${ctx.senderOpenId}`;
const peerId = isGroup ? (groupSession?.peerId ?? ctx.chatId) : ctx.senderOpenId;
const parentPeer = isGroup ? (groupSession?.parentPeer ?? null) : null;
const replyInThread = isGroup ? (groupSession?.replyInThread ?? false) : false;
if (isGroup && groupSession) {
log(
`feishu[${account.accountId}]: group session scope=${groupSession.groupSessionScope}, peer=${peerId}`,
);
}
let route = core.channel.routing.resolveAgentRoute({
cfg,
channel: "feishu",
accountId: account.accountId,
peer: {
refactor: unify peer kind to ChatType, rename dm to direct (#11881) * fix: use .js extension for ESM imports of RoutePeerKind The imports incorrectly used .ts extension which doesn't resolve with moduleResolution: NodeNext. Changed to .js and added 'type' import modifier. * fix tsconfig * refactor: unify peer kind to ChatType, rename dm to direct - Replace RoutePeerKind with ChatType throughout codebase - Change 'dm' literal values to 'direct' in routing/session keys - Keep backward compat: normalizeChatType accepts 'dm' -> 'direct' - Add ChatType export to plugin-sdk, deprecate RoutePeerKind - Update session key parsing to accept both 'dm' and 'direct' markers - Update all channel monitors and extensions to use ChatType BREAKING CHANGE: Session keys now use 'direct' instead of 'dm'. Existing 'dm' keys still work via backward compat layer. * fix tests * test: update session key expectations for dmdirect migration - Fix test expectations to expect :direct: in generated output - Add explicit backward compat test for normalizeChatType('dm') - Keep input test data with :dm: keys to verify backward compat * fix: accept legacy 'dm' in session key parsing for backward compat getDmHistoryLimitFromSessionKey now accepts both :dm: and :direct: to ensure old session keys continue to work correctly. * test: add explicit backward compat tests for dmdirect migration - session-key.test.ts: verify both :dm: and :direct: keys are valid - getDmHistoryLimitFromSessionKey: verify both formats work * feat: backward compat for resetByType.dm config key * test: skip unix-path Nix tests on Windows
2026-02-08 16:20:52 -08:00
kind: isGroup ? "group" : "direct",
id: peerId,
},
parentPeer,
});
// Dynamic agent creation for DM users
// When enabled, creates a unique agent instance with its own workspace for each DM user.
let effectiveCfg = cfg;
if (!isGroup && route.matchedBy === "default") {
const dynamicCfg = feishuCfg?.dynamicAgentCreation as DynamicAgentCreationConfig | undefined;
if (dynamicCfg?.enabled) {
const runtime = getFeishuRuntime();
const result = await maybeCreateDynamicAgent({
cfg,
runtime,
senderOpenId: ctx.senderOpenId,
dynamicCfg,
log: (msg) => log(msg),
});
if (result.created) {
effectiveCfg = result.updatedCfg;
// Re-resolve route with updated config
route = core.channel.routing.resolveAgentRoute({
cfg: result.updatedCfg,
channel: "feishu",
accountId: account.accountId,
peer: { kind: "direct", id: ctx.senderOpenId },
});
log(
`feishu[${account.accountId}]: dynamic agent created, new route: ${route.sessionKey}`,
);
}
}
}
const preview = ctx.content.replace(/\s+/g, " ").slice(0, 160);
const inboundLabel = isGroup
? `Feishu[${account.accountId}] message in group ${ctx.chatId}`
: `Feishu[${account.accountId}] DM from ${ctx.senderOpenId}`;
// Do not enqueue inbound user previews as system events.
// System events are prepended to future prompts and can be misread as
// authoritative transcript turns.
log(`feishu[${account.accountId}]: ${inboundLabel}: ${preview}`);
// Resolve media from message
const mediaMaxBytes = (feishuCfg?.mediaMaxMb ?? 30) * 1024 * 1024; // 30MB default
const mediaList = await resolveFeishuMediaList({
cfg,
messageId: ctx.messageId,
messageType: event.message.message_type,
content: event.message.content,
maxBytes: mediaMaxBytes,
log,
accountId: account.accountId,
});
const mediaPayload = buildAgentMediaPayload(mediaList);
// Fetch quoted/replied message content if parentId exists
let quotedContent: string | undefined;
if (ctx.parentId) {
try {
const quotedMsg = await getMessageFeishu({
cfg,
messageId: ctx.parentId,
accountId: account.accountId,
});
if (quotedMsg) {
quotedContent = quotedMsg.content;
log(
`feishu[${account.accountId}]: fetched quoted message: ${quotedContent?.slice(0, 100)}`,
);
}
} catch (err) {
log(`feishu[${account.accountId}]: failed to fetch quoted message: ${String(err)}`);
}
}
const envelopeOptions = core.channel.reply.resolveEnvelopeFormatOptions(cfg);
const messageBody = buildFeishuAgentBody({
ctx,
quotedContent,
permissionErrorForAgent,
fix(feishu): normalize all mentions in inbound agent context (#30252) * fix(feishu): normalize all mentions in inbound agent context Convert Feishu mention placeholders to explicit <at user_id="..."> tags (including bot mentions), add mention semantics hints for the model, and remove unused mentionMessageBody parsing to keep context handling consistent. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): use replacer callback and escape only < > in normalizeMentions Switch String.replace to a function replacer to prevent $ sequences in display names from being interpolated as replacement patterns. Narrow escaping to < and > only — & does not need escaping in LLM prompt tag bodies and escaping it degrades readability (e.g. R&D → R&amp;D). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): only use open_id in normalizeMentions tag, drop user_id fallback When a mention has no open_id, degrade to @name instead of emitting <at user_id="uid_...">. This keeps the tag user_id space exclusively open_id, so the bot self-reference hint (which uses botOpenId) is always consistent with what appears in the tags. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): register mention strip pattern for <at> tags in channel dock Add mentions.stripPatterns to feishuPlugin so that normalizeCommandBody receives a slash-clean string after normalizeMentions replaces Feishu placeholders with <at user_id="...">name</at> tags. Without this, group slash commands like @Bot /help had their leading / obscured by the tag prefix and no longer triggered command handlers. Pattern mirrors the approach used by Slack (<@[^>]+>) and Discord (<@!?\d+>). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * fix(feishu): strip bot mention in p2p to preserve DM slash commands In p2p messages the bot mention is a pure addressing prefix; converting it to <at user_id="..."> breaks slash commands because buildCommandContext skips stripMentions for DMs. Extend normalizeMentions with a stripKeys set and populate it with bot mention keys in p2p, so @Bot /help arrives as /help. Non-bot mentions (mention-forward targets) are still normalized to <at> tags in both p2p and group contexts. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> * Changelog: note Feishu inbound mention normalization --------- Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 12:40:17 +08:00
botOpenId,
});
const envelopeFrom = isGroup ? `${ctx.chatId}:${ctx.senderOpenId}` : ctx.senderOpenId;
if (permissionErrorForAgent) {
// Keep the notice in a single dispatch to avoid duplicate replies (#27372).
log(`feishu[${account.accountId}]: appending permission error notice to message body`);
}
const body = core.channel.reply.formatAgentEnvelope({
channel: "Feishu",
from: envelopeFrom,
timestamp: new Date(),
envelope: envelopeOptions,
body: messageBody,
});
let combinedBody = body;
const historyKey = groupHistoryKey;
if (isGroup && historyKey && chatHistories) {
combinedBody = buildPendingHistoryContextFromMap({
historyMap: chatHistories,
historyKey,
limit: historyLimit,
currentMessage: combinedBody,
formatEntry: (entry) =>
core.channel.reply.formatAgentEnvelope({
channel: "Feishu",
// Preserve speaker identity in group history as well.
from: `${ctx.chatId}:${entry.sender}`,
timestamp: entry.timestamp,
body: entry.body,
envelope: envelopeOptions,
}),
});
}
const inboundHistory =
isGroup && historyKey && historyLimit > 0 && chatHistories
? (chatHistories.get(historyKey) ?? []).map((entry) => ({
sender: entry.sender,
body: entry.body,
timestamp: entry.timestamp,
}))
: undefined;
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
// --- Shared context builder for dispatch ---
const buildCtxPayloadForAgent = (
agentSessionKey: string,
agentAccountId: string,
wasMentioned: boolean,
) =>
core.channel.reply.finalizeInboundContext({
Body: combinedBody,
BodyForAgent: messageBody,
InboundHistory: inboundHistory,
ReplyToId: ctx.parentId,
RootMessageId: ctx.rootId,
RawBody: ctx.content,
CommandBody: ctx.content,
From: feishuFrom,
To: feishuTo,
SessionKey: agentSessionKey,
AccountId: agentAccountId,
ChatType: isGroup ? "group" : "direct",
GroupSubject: isGroup ? ctx.chatId : undefined,
SenderName: ctx.senderName ?? ctx.senderOpenId,
SenderId: ctx.senderOpenId,
Provider: "feishu" as const,
Surface: "feishu" as const,
MessageSid: ctx.messageId,
ReplyToBody: quotedContent ?? undefined,
Timestamp: Date.now(),
WasMentioned: wasMentioned,
CommandAuthorized: commandAuthorized,
OriginatingChannel: "feishu" as const,
OriginatingTo: feishuTo,
GroupSystemPrompt: isGroup ? groupConfig?.systemPrompt?.trim() || undefined : undefined,
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
...mediaPayload,
});
// Parse message create_time (Feishu uses millisecond epoch string).
const messageCreateTimeMs = event.message.create_time
? parseInt(event.message.create_time, 10)
: undefined;
fix(feishu): comprehensive reply mechanism — outbound replyToId forwarding + topic-aware reply targeting (#33789) * fix(feishu): comprehensive reply mechanism fix — outbound replyToId forwarding + topic-aware reply targeting - Forward replyToId from ChannelOutboundContext through sendText/sendMedia to sendMessageFeishu/sendMarkdownCardFeishu/sendMediaFeishu, enabling reply-to-message via the message tool. - Fix group reply targeting: use ctx.messageId (triggering message) in normal groups to prevent silent topic thread creation (#32980). Preserve ctx.rootId targeting for topic-mode groups (group_topic/group_topic_sender) and groups with explicit replyInThread config. - Add regression tests for both fixes. Fixes #32980 Fixes #32958 Related #19784 * fix: normalize Feishu delivery.to before comparing with messaging tool targets - Add normalizeDeliveryTarget helper to strip user:/chat: prefixes for Feishu - Apply normalization in matchesMessagingToolDeliveryTarget before comparison - This ensures cron duplicate suppression works when session uses prefixed targets (user:ou_xxx) but messaging tool extract uses normalized bare IDs (ou_xxx) Fixes review comment on PR #32755 (cherry picked from commit fc20106f16ccc88a5f02e58922bb7b7999fe9dcd) * fix(feishu): catch thrown SDK errors for withdrawn reply targets The Feishu Lark SDK can throw exceptions (SDK errors with .code or AxiosErrors with .response.data.code) for withdrawn/deleted reply targets, in addition to returning error codes in the response object. Wrap reply calls in sendMessageFeishu and sendCardFeishu with try-catch to handle thrown withdrawn/not-found errors (230011, 231003) and fall back to client.im.message.create, matching the existing response-level fallback behavior. Also extract sendFallbackDirect helper to deduplicate the direct-send fallback block across both functions. Closes #33496 (cherry picked from commit ad0901aec103a2c52f186686cfaf5f8ba54b4a48) * feishu: forward outbound reply target context (cherry picked from commit c129a691fcf552a1cebe1e8a22ea8611ffc3b377) * feishu extension: tighten reply target fallback semantics (cherry picked from commit f85ec610f267020b66713c09e648ec004b2e26f1) * fix(feishu): align synthesized fallback typing and changelog attribution * test(feishu): cover group_topic_sender reply targeting --------- Co-authored-by: Xu Zimo <xuzimojimmy@163.com> Co-authored-by: Munem Hashmi <munem.hashmi@gmail.com> Co-authored-by: bmendonca3 <bmendonca3@users.noreply.github.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-05 10:32:28 +08:00
// Determine reply target based on group session mode:
// - Topic-mode groups (group_topic / group_topic_sender): reply to the topic
// root so the bot stays in the same thread.
// - Groups with explicit replyInThread config: reply to the root so the bot
// stays in the thread the user expects.
// - Normal groups (auto-detected threadReply from root_id): reply to the
// triggering message itself. Using rootId here would silently push the
// reply into a topic thread invisible in the main chat view (#32980).
const isTopicSession =
isGroup &&
(groupSession?.groupSessionScope === "group_topic" ||
groupSession?.groupSessionScope === "group_topic_sender");
const configReplyInThread =
isGroup &&
(groupConfig?.replyInThread ?? feishuCfg?.replyInThread ?? "disabled") === "enabled";
const replyTargetMessageId =
isTopicSession || configReplyInThread ? (ctx.rootId ?? ctx.messageId) : ctx.messageId;
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
const threadReply = isGroup ? (groupSession?.threadReply ?? false) : false;
if (broadcastAgents) {
// Cross-account dedup: in multi-account setups, Feishu delivers the same
// event to every bot account in the group. Only one account should handle
// broadcast dispatch to avoid duplicate agent sessions and race conditions.
// Uses a shared "broadcast" namespace (not per-account) so the first handler
// to reach this point claims the message; subsequent accounts skip.
if (!(await tryRecordMessagePersistent(ctx.messageId, "broadcast", log))) {
log(
`feishu[${account.accountId}]: broadcast already claimed by another account for message ${ctx.messageId}; skipping`,
);
return;
}
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
// --- Broadcast dispatch: send message to all configured agents ---
const strategy =
((cfg as Record<string, unknown>).broadcast as Record<string, unknown> | undefined)
?.strategy || "parallel";
const activeAgentId =
ctx.mentionedBot || !requireMention ? normalizeAgentId(route.agentId) : null;
const agentIds = (cfg.agents?.list ?? []).map((a: { id: string }) => normalizeAgentId(a.id));
const hasKnownAgents = agentIds.length > 0;
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
log(
`feishu[${account.accountId}]: broadcasting to ${broadcastAgents.length} agents (strategy=${strategy}, active=${activeAgentId ?? "none"})`,
);
const dispatchForAgent = async (agentId: string) => {
if (hasKnownAgents && !agentIds.includes(normalizeAgentId(agentId))) {
log(
`feishu[${account.accountId}]: broadcast agent ${agentId} not found in agents.list; skipping`,
);
return;
}
const agentSessionKey = buildBroadcastSessionKey(route.sessionKey, route.agentId, agentId);
const agentCtx = buildCtxPayloadForAgent(
agentSessionKey,
route.accountId,
ctx.mentionedBot && agentId === activeAgentId,
);
if (agentId === activeAgentId) {
// Active agent: real Feishu dispatcher (responds on Feishu)
const { dispatcher, replyOptions, markDispatchIdle } = createFeishuReplyDispatcher({
cfg,
agentId,
runtime: runtime as RuntimeEnv,
chatId: ctx.chatId,
replyToMessageId: replyTargetMessageId,
skipReplyToInMessages: !isGroup,
replyInThread,
rootId: ctx.rootId,
threadReply,
mentionTargets: ctx.mentionTargets,
accountId: account.accountId,
messageCreateTimeMs,
});
log(
`feishu[${account.accountId}]: broadcast active dispatch agent=${agentId} (session=${agentSessionKey})`,
);
await core.channel.reply.withReplyDispatcher({
dispatcher,
onSettled: () => markDispatchIdle(),
run: () =>
core.channel.reply.dispatchReplyFromConfig({
ctx: agentCtx,
cfg,
dispatcher,
replyOptions,
}),
});
} else {
// Observer agent: no-op dispatcher (session entry + inference, no Feishu reply).
// Strip CommandAuthorized so slash commands (e.g. /reset) don't silently
// mutate observer sessions — only the active agent should execute commands.
delete (agentCtx as Record<string, unknown>).CommandAuthorized;
const noopDispatcher = {
sendToolResult: () => false,
sendBlockReply: () => false,
sendFinalReply: () => false,
waitForIdle: async () => {},
getQueuedCounts: () => ({ tool: 0, block: 0, final: 0 }),
markComplete: () => {},
};
log(
`feishu[${account.accountId}]: broadcast observer dispatch agent=${agentId} (session=${agentSessionKey})`,
);
await core.channel.reply.withReplyDispatcher({
dispatcher: noopDispatcher,
run: () =>
core.channel.reply.dispatchReplyFromConfig({
ctx: agentCtx,
cfg,
dispatcher: noopDispatcher,
}),
});
}
};
if (strategy === "sequential") {
for (const agentId of broadcastAgents) {
try {
await dispatchForAgent(agentId);
} catch (err) {
log(
`feishu[${account.accountId}]: broadcast dispatch failed for agent=${agentId}: ${String(err)}`,
);
}
}
} else {
const results = await Promise.allSettled(broadcastAgents.map(dispatchForAgent));
for (let i = 0; i < results.length; i++) {
if (results[i].status === "rejected") {
log(
`feishu[${account.accountId}]: broadcast dispatch failed for agent=${broadcastAgents[i]}: ${String((results[i] as PromiseRejectedResult).reason)}`,
);
}
}
}
if (isGroup && historyKey && chatHistories) {
clearHistoryEntriesIfEnabled({
historyMap: chatHistories,
historyKey,
limit: historyLimit,
});
}
log(
`feishu[${account.accountId}]: broadcast dispatch complete for ${broadcastAgents.length} agents`,
);
} else {
// --- Single-agent dispatch (existing behavior) ---
const ctxPayload = buildCtxPayloadForAgent(
route.sessionKey,
route.accountId,
ctx.mentionedBot,
);
const { dispatcher, replyOptions, markDispatchIdle } = createFeishuReplyDispatcher({
cfg,
agentId: route.agentId,
runtime: runtime as RuntimeEnv,
chatId: ctx.chatId,
replyToMessageId: replyTargetMessageId,
skipReplyToInMessages: !isGroup,
replyInThread,
rootId: ctx.rootId,
threadReply,
mentionTargets: ctx.mentionTargets,
accountId: account.accountId,
messageCreateTimeMs,
});
feat(feishu): add broadcast support for multi-agent groups (#29575) * feat(feishu): add broadcast support for multi-agent group observation When multiple agents share a Feishu group chat, only the @mentioned agent receives the message. This prevents observer agents from building session memory of group activity they weren't directly addressed in. Adds broadcast support (reusing the same cfg.broadcast schema as WhatsApp) so all configured agents receive every group message in their session transcripts. Only the @mentioned agent responds on Feishu; observer agents process silently via no-op dispatchers. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): guard sequential broadcast dispatch against single-agent failure Wrap each dispatchForAgent() call in the sequential loop with try/catch so one agent's dispatch failure doesn't abort delivery to remaining agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): avoid duplicate messages in broadcast observer mode and normalize agent IDs - Skip recordPendingHistoryEntryIfEnabled for broadcast groups when not mentioned, since the message is dispatched directly to all agents. Previously the message appeared twice in the agent prompt. - Normalize agent IDs with toLowerCase() before membership checks so config casing mismatches don't silently skip valid agents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): set WasMentioned per-agent and normalize broadcast IDs - buildCtxPayloadForAgent now takes a wasMentioned parameter so active agents get WasMentioned=true and observers get false (P1 fix) - Normalize broadcastAgents to lowercase at resolution time and lowercase activeAgentId so all comparisons and session key generation use canonical IDs regardless of config casing (P2 fix) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): canonicalize broadcast agent IDs with normalizeAgentId * fix(feishu): match ReplyDispatcher sync return types for noop dispatcher The upstream ReplyDispatcher changed sendToolResult/sendBlockReply/ sendFinalReply to synchronous (returning boolean). Update the broadcast observer noop dispatcher to match. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): deduplicate broadcast agent IDs after normalization Config entries like "Main" and "main" collapse to the same canonical ID after normalizeAgentId but were dispatched multiple times. Use Set to deduplicate after normalization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): honor requireMention=false when selecting broadcast responder When requireMention is false, the routed agent should be active (reply on Feishu) even without an explicit @mention. Previously activeAgentId was null whenever ctx.mentionedBot was false, so all agents got the noop dispatcher and no reply was sent — silently breaking groups that disabled mention gating. Hoist requireMention out of the if(isGroup) block so it's accessible in the dispatch code. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): cross-account broadcast dedup to prevent duplicate dispatches In multi-account Feishu setups, the same message event is delivered to every bot account in a group. Without cross-account dedup, each account independently dispatches broadcast agents, causing 2×N dispatches instead of N (where N = number of broadcast agents). Two changes: 1. requireMention=true + bot not mentioned: return early instead of falling through to broadcast. The mentioned bot's handler will dispatch for all agents. Non-mentioned handlers record to history. 2. Add cross-account broadcast dedup using a shared 'broadcast' namespace (tryRecordMessagePersistent). The first handler to reach the broadcast block claims the message; subsequent accounts skip. This handles the requireMention=false multi-account case. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): strip CommandAuthorized from broadcast observer contexts Broadcast observer agents inherited CommandAuthorized from the sender, causing slash commands (e.g. /reset) to silently execute on every observer session. Now only the active agent retains CommandAuthorized; observers have it stripped before dispatch. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): use actual mention state for broadcast WasMentioned The active broadcast agent's WasMentioned was set to true whenever requireMention=false, even when the bot was not actually @mentioned. Now uses ctx.mentionedBot && agentId === activeAgentId, consistent with the single-agent path which passes ctx.mentionedBot directly. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(feishu): skip history buffer for broadcast accounts and log parallel failures 1. In requireMention groups with broadcast, non-mentioned accounts no longer buffer pending history — the mentioned handler's broadcast dispatch already writes turns into all agent sessions. Buffering caused duplicate replay via buildPendingHistoryContextFromMap. 2. Parallel broadcast dispatch now inspects Promise.allSettled results and logs rejected entries, matching the sequential path's per-agent error logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Changelog: note Feishu multi-agent broadcast dispatch * Changelog: restore author credit for Feishu broadcast entry --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Tak Hoffman <781889+Takhoffman@users.noreply.github.com>
2026-03-03 11:38:46 +08:00
log(`feishu[${account.accountId}]: dispatching to agent (session=${route.sessionKey})`);
const { queuedFinal, counts } = await core.channel.reply.withReplyDispatcher({
dispatcher,
onSettled: () => {
markDispatchIdle();
},
run: () =>
core.channel.reply.dispatchReplyFromConfig({
ctx: ctxPayload,
cfg,
dispatcher,
replyOptions,
}),
});
if (isGroup && historyKey && chatHistories) {
clearHistoryEntriesIfEnabled({
historyMap: chatHistories,
historyKey,
limit: historyLimit,
});
}
log(
`feishu[${account.accountId}]: dispatch complete (queuedFinal=${queuedFinal}, replies=${counts.final})`,
);
}
} catch (err) {
error(`feishu[${account.accountId}]: failed to dispatch message: ${String(err)}`);
}
}