fix(ollama): unify context window handling across discovery, merge, and OpenAI-compat transport (#29205)

* fix(ollama): inject num_ctx for OpenAI-compatible transport

* fix(ollama): discover per-model context and preserve higher limits

* fix(agents): prefer matching provider model for fallback limits

* fix(types): require numeric token limits in provider model merge

* fix(types): accept unknown payload in ollama num_ctx wrapper

* fix(types): simplify ollama settled-result extraction

* config(models): add provider flag for Ollama OpenAI num_ctx injection

* config(schema): allow provider num_ctx injection flag

* config(labels): label provider num_ctx injection flag

* config(help): document provider num_ctx injection flag

* agents(ollama): gate OpenAI num_ctx injection with provider config

* tests(ollama): cover provider num_ctx injection flag behavior

* docs(config): list provider num_ctx injection option

* docs(ollama): document OpenAI num_ctx injection toggle

* docs(config): clarify merge token-limit precedence

* config(help): note merge uses higher model token limits

* fix(ollama): cap /api/show discovery concurrency

* fix(ollama): restrict num_ctx injection to OpenAI compat

* tests(ollama): cover ipv6 and compat num_ctx gating

* fix(ollama): detect remote compat endpoints for ollama-labeled providers

* fix(ollama): cap per-model /api/show lookups to bound discovery load

src/agents/models-config.providers.ollama.test.ts

@@ -1,9 +1,14 @@
 import { mkdtempSync } from "node:fs";
 import { tmpdir } from "node:os";
 import { join } from "node:path";
-import { describe, expect, it } from "vitest";
+import { afterEach, describe, expect, it, vi } from "vitest";
 import { resolveImplicitProviders, resolveOllamaApiBase } from "./models-config.providers.js";
 
+afterEach(() => {
+  vi.unstubAllEnvs();
+  vi.unstubAllGlobals();
+});
+
 describe("resolveOllamaApiBase", () => {
   it("returns default localhost base when no configured URL is provided", () => {
     expect(resolveOllamaApiBase()).toBe("http://127.0.0.1:11434");
@@ -71,6 +76,110 @@ describe("Ollama provider", () => {
     }
   });
 
+  it("discovers per-model context windows from /api/show", async () => {
+    const agentDir = mkdtempSync(join(tmpdir(), "openclaw-test-"));
+    process.env.OLLAMA_API_KEY = "test-key";
+    vi.stubEnv("VITEST", "");
+    vi.stubEnv("NODE_ENV", "development");
+    const fetchMock = vi
+      .fn()
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({
+          models: [
+            { name: "qwen3:32b", modified_at: "", size: 1, digest: "" },
+            { name: "llama3.3:70b", modified_at: "", size: 1, digest: "" },
+          ],
+        }),
+      })
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({ model_info: { "qwen3.context_length": 131072 } }),
+      })
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({ model_info: { "llama.context_length": 65536 } }),
+      });
+    vi.stubGlobal("fetch", fetchMock);
+
+    try {
+      const providers = await resolveImplicitProviders({ agentDir });
+      const models = providers?.ollama?.models ?? [];
+      const qwen = models.find((model) => model.id === "qwen3:32b");
+      const llama = models.find((model) => model.id === "llama3.3:70b");
+      expect(qwen?.contextWindow).toBe(131072);
+      expect(llama?.contextWindow).toBe(65536);
+      expect(fetchMock).toHaveBeenCalledTimes(3);
+    } finally {
+      delete process.env.OLLAMA_API_KEY;
+    }
+  });
+
+  it("falls back to default context window when /api/show fails", async () => {
+    const agentDir = mkdtempSync(join(tmpdir(), "openclaw-test-"));
+    process.env.OLLAMA_API_KEY = "test-key";
+    vi.stubEnv("VITEST", "");
+    vi.stubEnv("NODE_ENV", "development");
+    const fetchMock = vi
+      .fn()
+      .mockResolvedValueOnce({
+        ok: true,
+        json: async () => ({
+          models: [{ name: "qwen3:32b", modified_at: "", size: 1, digest: "" }],
+        }),
+      })
+      .mockResolvedValueOnce({
+        ok: false,
+        status: 500,
+      });
+    vi.stubGlobal("fetch", fetchMock);
+
+    try {
+      const providers = await resolveImplicitProviders({ agentDir });
+      const model = providers?.ollama?.models?.find((entry) => entry.id === "qwen3:32b");
+      expect(model?.contextWindow).toBe(128000);
+      expect(fetchMock).toHaveBeenCalledTimes(2);
+    } finally {
+      delete process.env.OLLAMA_API_KEY;
+    }
+  });
+
+  it("caps /api/show requests when /api/tags returns a very large model list", async () => {
+    const agentDir = mkdtempSync(join(tmpdir(), "openclaw-test-"));
+    process.env.OLLAMA_API_KEY = "test-key";
+    vi.stubEnv("VITEST", "");
+    vi.stubEnv("NODE_ENV", "development");
+    const manyModels = Array.from({ length: 250 }, (_, idx) => ({
+      name: `model-${idx}`,
+      modified_at: "",
+      size: 1,
+      digest: "",
+    }));
+    const fetchMock = vi.fn(async (url: string) => {
+      if (url.endsWith("/api/tags")) {
+        return {
+          ok: true,
+          json: async () => ({ models: manyModels }),
+        };
+      }
+      return {
+        ok: true,
+        json: async () => ({ model_info: { "llama.context_length": 65536 } }),
+      };
+    });
+    vi.stubGlobal("fetch", fetchMock);
+
+    try {
+      const providers = await resolveImplicitProviders({ agentDir });
+      const models = providers?.ollama?.models ?? [];
+      // 1 call for /api/tags + 200 capped /api/show calls.
+      expect(fetchMock).toHaveBeenCalledTimes(201);
+      expect(models).toHaveLength(200);
+    } finally {
+      delete process.env.OLLAMA_API_KEY;
+    }
+  });
+
   it("should have correct model structure without streaming override", () => {
     const mockOllamaModel = {
       id: "llama3.3:latest",