fix(provider): onboard azure custom endpoints via responses
This commit is contained in:
Ayaan Zaidi
parent
9d772d6eab
commit
7a57082466
@ -44,7 +44,6 @@ Docs: https://docs.openclaw.ai
|
||||
- Control UI/chat: add an expand-to-canvas button on assistant chat bubbles and in-app session navigation from Sessions and Cron views. Thanks @BunsDev.
|
||||
- Plugins/context engines: expose `delegateCompactionToRuntime(...)` on the public plugin SDK, refactor the legacy engine to use the shared helper, and clarify `ownsCompaction` delegation semantics for non-owning engines. (#49061) Thanks @jalehman.
|
||||
- Plugins/MiniMax: add MiniMax-M2.7 and MiniMax-M2.7-highspeed models and update the default model from M2.5 to M2.7. (#49691) Thanks @liyuan97.
|
||||
- Contracts/Matrix: validate Matrix session binding coverage through the real manager, expose the manager on the Matrix runtime API, and let tests pass an explicit state directory for isolated contract setup. (#50369) thanks @ChroniCat.
|
||||
|
||||
### Fixes
|
||||
|
||||
@ -93,6 +92,7 @@ Docs: https://docs.openclaw.ai
|
||||
- Z.AI/onboarding: add `glm-5-turbo` to the default Z.AI provider catalog so onboarding-generated configs expose the new model alongside the existing GLM defaults. (#46670) Thanks @tomsun28.
|
||||
- Zalo Personal/group gating: stop reapplying `dmPolicy.allowFrom` as a sender gate for already-allowlisted groups when `groupAllowFrom` is unset, so any member of an allowed group can trigger replies while DMs stay restricted. (#46663) Fixes #40146. Thanks @Takhoffman.
|
||||
- Zalo/plugin runtime: export `resolveClientIp` from `openclaw/plugin-sdk/zalo` so installed builds no longer crash on startup when the webhook monitor loads from the packaged extension instead of the monorepo source tree. (#46549) Thanks @No898.
|
||||
- Onboarding/custom providers: store Azure OpenAI and Azure AI Foundry custom endpoints with the Responses API config shape, normalized `/openai/v1` base URLs, and Azure-safe defaults so TUI and agent runs work after setup. (#49543) Thanks @kunalk16.
|
||||
- Docker/live tests: mount external CLI auth homes into writable container copies, derive Codex OAuth expiry from JWT `exp`, refresh synced CLI creds instead of trusting stale cached expiry, and make gateway live probes wait on transcript output so `pnpm test:docker:all` stays green in Linux.
|
||||
- Plugins/install precedence: keep bundled plugins ahead of auto-discovered globals by default, but let an explicitly installed plugin record win its own duplicate-id tie so installed channel plugins load from `~/.openclaw/extensions` after `openclaw plugins install`. (#46722) Thanks @Takhoffman.
|
||||
- Control UI/logging: make browser-safe logger imports avoid eager temp-dir resolution so the bundled Control UI no longer crashes to a blank screen when logging reaches `tmp-openclaw-dir`. (#48469) Fixes #48062. Thanks @7inspire.
|
||||
|
||||
@ -188,7 +188,7 @@ describe("promptCustomApiConfig", () => {
|
||||
expect(JSON.parse(firstCall?.body ?? "{}")).toMatchObject({ max_tokens: 1 });
|
||||
});
|
||||
|
||||
it("uses azure-specific headers and body for openai verification probes", async () => {
|
||||
it("uses azure responses-specific headers and body for openai verification probes", async () => {
|
||||
const prompter = createTestPrompter({
|
||||
text: [
|
||||
"https://my-resource.openai.azure.com",
|
||||
@ -213,18 +213,16 @@ describe("promptCustomApiConfig", () => {
|
||||
}
|
||||
const parsedBody = JSON.parse(firstInit?.body ?? "{}");
|
||||
|
||||
expect(firstUrl).toContain("/openai/deployments/gpt-4.1/chat/completions");
|
||||
expect(firstUrl).toContain("api-version=2024-10-21");
|
||||
expect(firstUrl).toBe("https://my-resource.openai.azure.com/openai/v1/responses");
|
||||
expect(firstInit?.headers?.["api-key"]).toBe("azure-test-key");
|
||||
expect(firstInit?.headers?.Authorization).toBeUndefined();
|
||||
expect(firstInit?.body).toBeDefined();
|
||||
expect(parsedBody).toMatchObject({
|
||||
messages: [{ role: "user", content: "Hi" }],
|
||||
max_completion_tokens: 5,
|
||||
expect(parsedBody).toEqual({
|
||||
model: "gpt-4.1",
|
||||
input: "Hi",
|
||||
max_output_tokens: 1,
|
||||
stream: false,
|
||||
});
|
||||
expect(parsedBody).not.toHaveProperty("model");
|
||||
expect(parsedBody).not.toHaveProperty("max_tokens");
|
||||
});
|
||||
|
||||
it("uses expanded max_tokens for anthropic verification probes", async () => {
|
||||
@ -432,6 +430,192 @@ describe("applyCustomApiConfig", () => {
|
||||
])("rejects $name", ({ params, expectedMessage }) => {
|
||||
expect(() => applyCustomApiConfig(params)).toThrow(expectedMessage);
|
||||
});
|
||||
|
||||
it("produces azure-specific config for Azure OpenAI URLs with reasoning model", () => {
|
||||
const result = applyCustomApiConfig({
|
||||
config: {},
|
||||
baseUrl: "https://user123-resource.openai.azure.com",
|
||||
modelId: "o4-mini",
|
||||
compatibility: "openai",
|
||||
apiKey: "abcd1234",
|
||||
});
|
||||
const providerId = result.providerId!;
|
||||
const provider = result.config.models?.providers?.[providerId];
|
||||
|
||||
expect(provider?.baseUrl).toBe("https://user123-resource.openai.azure.com/openai/v1");
|
||||
expect(provider?.api).toBe("openai-responses");
|
||||
expect(provider?.authHeader).toBe(false);
|
||||
expect(provider?.headers).toEqual({ "api-key": "abcd1234" });
|
||||
|
||||
const model = provider?.models?.find((m) => m.id === "o4-mini");
|
||||
expect(model?.input).toEqual(["text", "image"]);
|
||||
expect(model?.reasoning).toBe(true);
|
||||
expect(model?.compat).toEqual({ supportsStore: false });
|
||||
|
||||
const modelRef = `${providerId}/${result.modelId}`;
|
||||
expect(result.config.agents?.defaults?.models?.[modelRef]?.params?.thinking).toBe("medium");
|
||||
});
|
||||
|
||||
it("produces azure-specific config for Azure AI Foundry URLs", () => {
|
||||
const result = applyCustomApiConfig({
|
||||
config: {},
|
||||
baseUrl: "https://my-resource.services.ai.azure.com",
|
||||
modelId: "gpt-4.1",
|
||||
compatibility: "openai",
|
||||
apiKey: "key123",
|
||||
});
|
||||
const providerId = result.providerId!;
|
||||
const provider = result.config.models?.providers?.[providerId];
|
||||
|
||||
expect(provider?.baseUrl).toBe("https://my-resource.services.ai.azure.com/openai/v1");
|
||||
expect(provider?.api).toBe("openai-responses");
|
||||
expect(provider?.authHeader).toBe(false);
|
||||
expect(provider?.headers).toEqual({ "api-key": "key123" });
|
||||
|
||||
const model = provider?.models?.find((m) => m.id === "gpt-4.1");
|
||||
expect(model?.reasoning).toBe(false);
|
||||
expect(model?.input).toEqual(["text"]);
|
||||
expect(model?.compat).toEqual({ supportsStore: false });
|
||||
|
||||
const modelRef = `${providerId}/gpt-4.1`;
|
||||
expect(result.config.agents?.defaults?.models?.[modelRef]?.params?.thinking).toBeUndefined();
|
||||
});
|
||||
|
||||
it("strips pre-existing deployment path from Azure URL in stored config", () => {
|
||||
const result = applyCustomApiConfig({
|
||||
config: {},
|
||||
baseUrl: "https://my-resource.openai.azure.com/openai/deployments/gpt-4",
|
||||
modelId: "gpt-4",
|
||||
compatibility: "openai",
|
||||
apiKey: "key456",
|
||||
});
|
||||
const providerId = result.providerId!;
|
||||
const provider = result.config.models?.providers?.[providerId];
|
||||
|
||||
expect(provider?.baseUrl).toBe("https://my-resource.openai.azure.com/openai/v1");
|
||||
});
|
||||
|
||||
it("re-onboard updates existing Azure provider instead of creating a duplicate", () => {
|
||||
const oldProviderId = "custom-my-resource-openai-azure-com";
|
||||
const result = applyCustomApiConfig({
|
||||
config: {
|
||||
models: {
|
||||
providers: {
|
||||
[oldProviderId]: {
|
||||
baseUrl: "https://my-resource.openai.azure.com/openai/deployments/gpt-4",
|
||||
api: "openai-completions",
|
||||
models: [
|
||||
{
|
||||
id: "gpt-4",
|
||||
name: "gpt-4",
|
||||
contextWindow: 1,
|
||||
maxTokens: 1,
|
||||
input: ["text"],
|
||||
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
|
||||
reasoning: false,
|
||||
},
|
||||
],
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
baseUrl: "https://my-resource.openai.azure.com",
|
||||
modelId: "gpt-4",
|
||||
compatibility: "openai",
|
||||
apiKey: "key789",
|
||||
});
|
||||
|
||||
expect(result.providerId).toBe(oldProviderId);
|
||||
expect(result.providerIdRenamedFrom).toBeUndefined();
|
||||
const provider = result.config.models?.providers?.[oldProviderId];
|
||||
expect(provider?.baseUrl).toBe("https://my-resource.openai.azure.com/openai/v1");
|
||||
expect(provider?.api).toBe("openai-responses");
|
||||
expect(provider?.authHeader).toBe(false);
|
||||
expect(provider?.headers).toEqual({ "api-key": "key789" });
|
||||
});
|
||||
|
||||
it("does not add azure fields for non-azure URLs", () => {
|
||||
const result = applyCustomApiConfig({
|
||||
config: {},
|
||||
baseUrl: "https://llm.example.com/v1",
|
||||
modelId: "foo-large",
|
||||
compatibility: "openai",
|
||||
apiKey: "key123",
|
||||
providerId: "custom",
|
||||
});
|
||||
const provider = result.config.models?.providers?.custom;
|
||||
|
||||
expect(provider?.api).toBe("openai-completions");
|
||||
expect(provider?.authHeader).toBeUndefined();
|
||||
expect(provider?.headers).toBeUndefined();
|
||||
expect(provider?.models?.[0]?.reasoning).toBe(false);
|
||||
expect(provider?.models?.[0]?.input).toEqual(["text"]);
|
||||
expect(provider?.models?.[0]?.compat).toBeUndefined();
|
||||
expect(
|
||||
result.config.agents?.defaults?.models?.["custom/foo-large"]?.params?.thinking,
|
||||
).toBeUndefined();
|
||||
});
|
||||
|
||||
it("re-onboard preserves user-customized fields for non-azure models", () => {
|
||||
const result = applyCustomApiConfig({
|
||||
config: {
|
||||
models: {
|
||||
providers: {
|
||||
custom: {
|
||||
baseUrl: "https://llm.example.com/v1",
|
||||
api: "openai-completions",
|
||||
models: [
|
||||
{
|
||||
id: "foo-large",
|
||||
name: "My Custom Model",
|
||||
reasoning: true,
|
||||
input: ["text", "image"],
|
||||
cost: { input: 1, output: 2, cacheRead: 0, cacheWrite: 0 },
|
||||
contextWindow: 131072,
|
||||
maxTokens: 16384,
|
||||
},
|
||||
],
|
||||
},
|
||||
},
|
||||
},
|
||||
} as OpenClawConfig,
|
||||
baseUrl: "https://llm.example.com/v1",
|
||||
modelId: "foo-large",
|
||||
compatibility: "openai",
|
||||
apiKey: "key",
|
||||
providerId: "custom",
|
||||
});
|
||||
const model = result.config.models?.providers?.custom?.models?.find(
|
||||
(m) => m.id === "foo-large",
|
||||
);
|
||||
expect(model?.name).toBe("My Custom Model");
|
||||
expect(model?.reasoning).toBe(true);
|
||||
expect(model?.input).toEqual(["text", "image"]);
|
||||
expect(model?.cost).toEqual({ input: 1, output: 2, cacheRead: 0, cacheWrite: 0 });
|
||||
expect(model?.maxTokens).toBe(16384);
|
||||
expect(model?.contextWindow).toBe(131072);
|
||||
});
|
||||
|
||||
it("preserves existing per-model thinking when already set for azure reasoning model", () => {
|
||||
const providerId = "custom-my-resource-openai-azure-com";
|
||||
const modelRef = `${providerId}/o3-mini`;
|
||||
const result = applyCustomApiConfig({
|
||||
config: {
|
||||
agents: {
|
||||
defaults: {
|
||||
models: {
|
||||
[modelRef]: { params: { thinking: "high" } },
|
||||
},
|
||||
},
|
||||
},
|
||||
} as OpenClawConfig,
|
||||
baseUrl: "https://my-resource.openai.azure.com",
|
||||
modelId: "o3-mini",
|
||||
compatibility: "openai",
|
||||
apiKey: "key",
|
||||
});
|
||||
expect(result.config.agents?.defaults?.models?.[modelRef]?.params?.thinking).toBe("high");
|
||||
});
|
||||
});
|
||||
|
||||
describe("parseNonInteractiveCustomApiFlags", () => {
|
||||
|
||||
@ -19,6 +19,9 @@ import type { SecretInputMode } from "./onboard-types.js";
|
||||
|
||||
const DEFAULT_CONTEXT_WINDOW = CONTEXT_WINDOW_HARD_MIN_TOKENS;
|
||||
const DEFAULT_MAX_TOKENS = 4096;
|
||||
// Azure OpenAI uses the Responses API which supports larger defaults
|
||||
const AZURE_DEFAULT_CONTEXT_WINDOW = 400_000;
|
||||
const AZURE_DEFAULT_MAX_TOKENS = 16_384;
|
||||
const VERIFY_TIMEOUT_MS = 30_000;
|
||||
|
||||
function normalizeContextWindowForCustomModel(value: unknown): number {
|
||||
@ -61,6 +64,32 @@ function transformAzureUrl(baseUrl: string, modelId: string): string {
|
||||
return `${normalizedUrl}/openai/deployments/${modelId}`;
|
||||
}
|
||||
|
||||
/**
|
||||
* Transforms an Azure URL into the base URL stored in config.
|
||||
*
|
||||
* Example:
|
||||
* https://my-resource.openai.azure.com
|
||||
* => https://my-resource.openai.azure.com/openai/v1
|
||||
*/
|
||||
function transformAzureConfigUrl(baseUrl: string): string {
|
||||
const normalizedUrl = baseUrl.endsWith("/") ? baseUrl.slice(0, -1) : baseUrl;
|
||||
if (normalizedUrl.endsWith("/openai/v1")) {
|
||||
return normalizedUrl;
|
||||
}
|
||||
// Strip a full deployment path back to the base origin
|
||||
const deploymentIdx = normalizedUrl.indexOf("/openai/deployments/");
|
||||
const base = deploymentIdx !== -1 ? normalizedUrl.slice(0, deploymentIdx) : normalizedUrl;
|
||||
return `${base}/openai/v1`;
|
||||
}
|
||||
|
||||
function hasSameHost(a: string, b: string): boolean {
|
||||
try {
|
||||
return new URL(a).hostname.toLowerCase() === new URL(b).hostname.toLowerCase();
|
||||
} catch {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
export type CustomApiCompatibility = "openai" | "anthropic";
|
||||
type CustomApiCompatibilityChoice = CustomApiCompatibility | "unknown";
|
||||
export type CustomApiResult = {
|
||||
@ -174,7 +203,11 @@ function resolveUniqueEndpointId(params: {
|
||||
}) {
|
||||
const normalized = normalizeEndpointId(params.requestedId) || "custom";
|
||||
const existing = params.providers[normalized];
|
||||
if (!existing?.baseUrl || existing.baseUrl === params.baseUrl) {
|
||||
if (
|
||||
!existing?.baseUrl ||
|
||||
existing.baseUrl === params.baseUrl ||
|
||||
(isAzureUrl(params.baseUrl) && hasSameHost(existing.baseUrl, params.baseUrl))
|
||||
) {
|
||||
return { providerId: normalized, renamed: false };
|
||||
}
|
||||
let suffix = 2;
|
||||
@ -320,26 +353,31 @@ async function requestOpenAiVerification(params: {
|
||||
apiKey: string;
|
||||
modelId: string;
|
||||
}): Promise<VerificationResult> {
|
||||
const endpoint = resolveVerificationEndpoint({
|
||||
baseUrl: params.baseUrl,
|
||||
modelId: params.modelId,
|
||||
endpointPath: "chat/completions",
|
||||
});
|
||||
const isBaseUrlAzureUrl = isAzureUrl(params.baseUrl);
|
||||
const headers = isBaseUrlAzureUrl
|
||||
? buildAzureOpenAiHeaders(params.apiKey)
|
||||
: buildOpenAiHeaders(params.apiKey);
|
||||
if (isBaseUrlAzureUrl) {
|
||||
const endpoint = new URL(
|
||||
"responses",
|
||||
transformAzureConfigUrl(params.baseUrl).replace(/\/?$/, "/"),
|
||||
).href;
|
||||
return await requestVerification({
|
||||
endpoint,
|
||||
headers,
|
||||
body: {
|
||||
messages: [{ role: "user", content: "Hi" }],
|
||||
max_completion_tokens: 5,
|
||||
model: params.modelId,
|
||||
input: "Hi",
|
||||
max_output_tokens: 1,
|
||||
stream: false,
|
||||
},
|
||||
});
|
||||
} else {
|
||||
const endpoint = resolveVerificationEndpoint({
|
||||
baseUrl: params.baseUrl,
|
||||
modelId: params.modelId,
|
||||
endpointPath: "chat/completions",
|
||||
});
|
||||
return await requestVerification({
|
||||
endpoint,
|
||||
headers,
|
||||
@ -572,8 +610,9 @@ export function applyCustomApiConfig(params: ApplyCustomApiConfigParams): Custom
|
||||
throw new CustomApiError("invalid_model_id", "Custom provider model ID is required.");
|
||||
}
|
||||
|
||||
const isAzure = isAzureUrl(baseUrl);
|
||||
// Transform Azure URLs to include the deployment path for API calls
|
||||
const resolvedBaseUrl = isAzureUrl(baseUrl) ? transformAzureUrl(baseUrl, modelId) : baseUrl;
|
||||
const resolvedBaseUrl = isAzure ? transformAzureConfigUrl(baseUrl) : baseUrl;
|
||||
|
||||
const providerIdResult = resolveCustomProviderId({
|
||||
config: params.config,
|
||||
@ -597,21 +636,39 @@ export function applyCustomApiConfig(params: ApplyCustomApiConfigParams): Custom
|
||||
const existingProvider = providers[providerId];
|
||||
const existingModels = Array.isArray(existingProvider?.models) ? existingProvider.models : [];
|
||||
const hasModel = existingModels.some((model) => model.id === modelId);
|
||||
const nextModel = {
|
||||
id: modelId,
|
||||
name: `${modelId} (Custom Provider)`,
|
||||
contextWindow: DEFAULT_CONTEXT_WINDOW,
|
||||
maxTokens: DEFAULT_MAX_TOKENS,
|
||||
input: ["text"] as ["text"],
|
||||
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
|
||||
reasoning: false,
|
||||
};
|
||||
const isLikelyReasoningModel = isAzure && /\b(o[134]|gpt-([5-9]|\d{2,}))\b/i.test(modelId);
|
||||
const nextModel = isAzure
|
||||
? {
|
||||
id: modelId,
|
||||
name: `${modelId} (Custom Provider)`,
|
||||
contextWindow: AZURE_DEFAULT_CONTEXT_WINDOW,
|
||||
maxTokens: AZURE_DEFAULT_MAX_TOKENS,
|
||||
input: isLikelyReasoningModel
|
||||
? (["text", "image"] as Array<"text" | "image">)
|
||||
: (["text"] as ["text"]),
|
||||
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
|
||||
reasoning: isLikelyReasoningModel,
|
||||
compat: { supportsStore: false },
|
||||
}
|
||||
: {
|
||||
id: modelId,
|
||||
name: `${modelId} (Custom Provider)`,
|
||||
contextWindow: DEFAULT_CONTEXT_WINDOW,
|
||||
maxTokens: DEFAULT_MAX_TOKENS,
|
||||
input: ["text"] as ["text"],
|
||||
cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
|
||||
reasoning: false,
|
||||
};
|
||||
const mergedModels = hasModel
|
||||
? existingModels.map((model) =>
|
||||
model.id === modelId
|
||||
? {
|
||||
...model,
|
||||
...(isAzure ? nextModel : {}),
|
||||
name: model.name ?? nextModel.name,
|
||||
cost: model.cost ?? nextModel.cost,
|
||||
contextWindow: normalizeContextWindowForCustomModel(model.contextWindow),
|
||||
maxTokens: model.maxTokens ?? nextModel.maxTokens,
|
||||
}
|
||||
: model,
|
||||
)
|
||||
@ -621,6 +678,11 @@ export function applyCustomApiConfig(params: ApplyCustomApiConfigParams): Custom
|
||||
normalizeOptionalProviderApiKey(params.apiKey) ??
|
||||
normalizeOptionalProviderApiKey(existingApiKey);
|
||||
|
||||
const providerApi = isAzure
|
||||
? ("openai-responses" as const)
|
||||
: resolveProviderApi(params.compatibility);
|
||||
const azureHeaders = isAzure && normalizedApiKey ? { "api-key": normalizedApiKey } : undefined;
|
||||
|
||||
let config: OpenClawConfig = {
|
||||
...params.config,
|
||||
models: {
|
||||
@ -631,8 +693,10 @@ export function applyCustomApiConfig(params: ApplyCustomApiConfigParams): Custom
|
||||
[providerId]: {
|
||||
...existingProviderRest,
|
||||
baseUrl: resolvedBaseUrl,
|
||||
api: resolveProviderApi(params.compatibility),
|
||||
api: providerApi,
|
||||
...(normalizedApiKey ? { apiKey: normalizedApiKey } : {}),
|
||||
...(isAzure ? { authHeader: false } : {}),
|
||||
...(azureHeaders ? { headers: azureHeaders } : {}),
|
||||
models: mergedModels.length > 0 ? mergedModels : [nextModel],
|
||||
},
|
||||
},
|
||||
@ -640,6 +704,30 @@ export function applyCustomApiConfig(params: ApplyCustomApiConfigParams): Custom
|
||||
};
|
||||
|
||||
config = applyPrimaryModel(config, modelRef);
|
||||
if (isAzure && isLikelyReasoningModel) {
|
||||
const existingPerModelThinking = config.agents?.defaults?.models?.[modelRef]?.params?.thinking;
|
||||
if (!existingPerModelThinking) {
|
||||
config = {
|
||||
...config,
|
||||
agents: {
|
||||
...config.agents,
|
||||
defaults: {
|
||||
...config.agents?.defaults,
|
||||
models: {
|
||||
...config.agents?.defaults?.models,
|
||||
[modelRef]: {
|
||||
...config.agents?.defaults?.models?.[modelRef],
|
||||
params: {
|
||||
...config.agents?.defaults?.models?.[modelRef]?.params,
|
||||
thinking: "medium",
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
},
|
||||
};
|
||||
}
|
||||
}
|
||||
if (alias) {
|
||||
config = {
|
||||
...config,
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user