fix(copilot): keep Gemini working without live discovery (#131143)

Declare the Gemini Chat Completions transport and request compatibility in the owning model manifest. Remove the private metadata decoration loop so static catalog selection and forward-compatible synthesis use the same declared contract.

docs/providers/github-copilot.md

@@ -289,8 +289,9 @@ configured default model is never replaced.
   <Accordion title="Transport selection">
     Claude model IDs use the Anthropic Messages transport automatically.
     Gemini models use the OpenAI Chat Completions transport; GPT and o-series
-    models keep the OpenAI Responses transport. OpenClaw selects the correct
-    transport based on the model ref.
+    models keep the OpenAI Responses transport. The bundled static catalog
+    includes these transports and request compatibility settings, so Gemini
+    keeps using Chat Completions when live discovery is disabled or unavailable.
   </Accordion>
 
   <Accordion title="Thinking levels">

extensions/github-copilot/model-metadata.ts

@@ -26,10 +26,6 @@ const manifestModels = buildManifestModelProviderConfig({
   providerId: "github-copilot",
   catalog: manifestCatalog,
 }).models;
-for (const model of manifestModels) {
-  model.api = resolveCopilotTransportApi(model.id);
-  model.compat = { ...resolveCopilotModelCompat(model.id), ...model.compat };
-}
 
 const STATIC_MODEL_OVERRIDES = new Map<string, Partial<ModelDefinitionConfig>>([
   ...manifestModels.map((model) => [model.id, model] as const),

extensions/github-copilot/openclaw.plugin.json

@@ -97,6 +97,13 @@
           {
             "id": "gemini-3.6-flash",
             "name": "Gemini 3.6 Flash",
+            "api": "openai-completions",
+            "compat": {
+              "supportsStore": false,
+              "supportsDeveloperRole": false,
+              "supportsUsageInStreaming": false,
+              "maxTokensField": "max_tokens"
+            },
             "input": ["text", "image"],
             "contextWindow": 1048576,
             "maxTokens": 65536,
@@ -105,6 +112,13 @@
           {
             "id": "gemini-3.1-pro-preview",
             "name": "Gemini 3.1 Pro Preview",
+            "api": "openai-completions",
+            "compat": {
+              "supportsStore": false,
+              "supportsDeveloperRole": false,
+              "supportsUsageInStreaming": false,
+              "maxTokensField": "max_tokens"
+            },
             "input": ["text", "image"],
             "contextWindow": 1048576,
             "maxTokens": 65536,
@@ -113,6 +127,13 @@
           {
             "id": "gemini-3.5-flash",
             "name": "Gemini 3.5 Flash",
+            "api": "openai-completions",
+            "compat": {
+              "supportsStore": false,
+              "supportsDeveloperRole": false,
+              "supportsUsageInStreaming": false,
+              "maxTokensField": "max_tokens"
+            },
             "status": "deprecated",
             "replacedBy": "gemini-3.6-flash",
             "input": ["text", "image"],
@@ -123,6 +144,13 @@
           {
             "id": "gemini-2.5-pro",
             "name": "Gemini 2.5 Pro",
+            "api": "openai-completions",
+            "compat": {
+              "supportsStore": false,
+              "supportsDeveloperRole": false,
+              "supportsUsageInStreaming": false,
+              "maxTokensField": "max_tokens"
+            },
             "status": "deprecated",
             "replacedBy": "gemini-3.1-pro-preview",
             "input": ["text", "image"],

src/plugin-sdk/test-helpers/provider-runtime-contract.ts

@@ -1,5 +1,7 @@
 // Provider runtime contract helpers define reusable runtime tests for provider plugins.
+import { normalizeModelCatalog } from "@openclaw/model-catalog-core/model-catalog-normalize";
 import { afterAll, beforeAll, beforeEach, describe, expect, it, vi } from "vitest";
+import { createPluginMetadataSnapshot } from "../../config/plugin-auto-enable.test-helpers.js";
 import type { ProviderRuntimeModel } from "../plugin-entry.js";
 import { registerProviderPlugin, requireRegisteredProvider } from "../plugin-test-runtime.js";
 import { buildManifestModelProviderConfig } from "../provider-catalog-shared.js";
@@ -283,6 +285,95 @@ export function describeGithubCopilotProviderRuntimeContract(
       ]);
       const createManifestModel = createManifestModelFactory("github-copilot", manifestCatalog);
 
+      it.each([
+        ["gemini-3.6-flash", "openai-completions", false],
+        ["gemini-3.1-pro-preview", "openai-completions", false],
+        ["gemini-3.5-flash", "openai-completions", false],
+        ["gemini-2.5-pro", "openai-completions", false],
+        ["gpt-5.6-sol", "openai-responses", false],
+        ["claude-sonnet-5", "anthropic-messages", false],
+        ["gemini-3.6-flash", "openai-completions", true],
+      ] as const)(
+        "routes static %s through %s with discovery enabled=%s",
+        async (modelId, api, discoveryEnabled) => {
+          const { createBundledStaticCatalogModelResolver } =
+            await import("../../agents/embedded-agent-runner/model.static-catalog.js");
+          const config = {
+            models: { catalogRefresh: { enabled: false } },
+            plugins: {
+              entries: {
+                "github-copilot": { config: { discovery: { enabled: discoveryEnabled } } },
+              },
+            },
+          };
+          const metadataSnapshot = createPluginMetadataSnapshot({
+            config,
+            manifestRegistry: {
+              diagnostics: [],
+              plugins: [
+                {
+                  id: "github-copilot",
+                  origin: "bundled",
+                  providers: ["github-copilot"],
+                  channels: [],
+                  cliBackends: [],
+                  skills: [],
+                  hooks: [],
+                  rootDir: "/fixtures/github-copilot",
+                  source: "/fixtures/github-copilot/index.js",
+                  manifestPath: "/fixtures/github-copilot/openclaw.plugin.json",
+                  modelCatalog: normalizeModelCatalog(
+                    {
+                      providers: { "github-copilot": manifestCatalog },
+                      discovery: { "github-copilot": "runtime" },
+                    },
+                    { ownedProviders: new Set(["github-copilot"]) },
+                  ),
+                },
+              ],
+            },
+          });
+          const resolveModel = createBundledStaticCatalogModelResolver({
+            cfg: config,
+            env: {},
+            metadataSnapshot,
+            includeRuntimeDiscovery: true,
+          });
+          const model = resolveModel({ provider: "github-copilot", modelId });
+          expect(model?.api).toBe(api);
+          if (api === "openai-completions") {
+            expect(model?.compat).toMatchObject({
+              supportsStore: false,
+              supportsDeveloperRole: false,
+              supportsUsageInStreaming: false,
+              maxTokensField: "max_tokens",
+            });
+          }
+          const provider = requireProviderContractProvider("github-copilot");
+          expect(
+            provider.preferRuntimeResolvedModel?.({
+              config,
+              provider: "github-copilot",
+              modelId,
+            }),
+          ).toBe(discoveryEnabled);
+          // A missing prepared live row also occurs when enabled discovery is unavailable.
+          expect(
+            provider.resolveDynamicModel?.({
+              config,
+              provider: "github-copilot",
+              modelId,
+              modelRegistry: {
+                find: () => model,
+                getAll: () => (model ? [model] : []),
+                getAvailable: () => (model ? [model] : []),
+                hasConfiguredAuth: () => false,
+              },
+            }),
+          ).toBeUndefined();
+        },
+      );
+
       it("owns Copilot-specific forward-compat fallbacks", () => {
         const provider = requireProviderContractProvider("github-copilot");
         const expected = createManifestModel("gpt-5.4");