feat(KB-025): add per-task model overrides for executor and validator

- Add model override fields to Task type and TaskStore (modelProvider, modelId, validatorModelProvider, validatorModelId) - Extend PATCH /api/tasks/:id endpoint with validation for model fields - Create ModelSelectorTab component with provider/model dropdowns and tests - Integrate Model tab into TaskDetailModal for per-task model configuration - Update executor to use per-task model overrides when both provider and modelId are set - Update reviewer to use per-task validator model overrides in reviewStep - Document per-task model selection feature in AGENTS.md settings section
2026-03-29 19:36:15 -07:00
parent 2c865d9914
commit a5393463d5
16 changed files with 1157 additions and 13 deletions
--- a/packages/engine/src/executor.test.ts
+++ b/packages/engine/src/executor.test.ts
@@ -2852,3 +2852,176 @@ describe("TaskExecutor usage limit detection", () => {
    );
  });
 });
+
+describe("Per-task model overrides", () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+    mockedExistsSync.mockReturnValue(true);
+  });
+
+  it("uses per-task model overrides when both provider and modelId are set", async () => {
+    const store = createMockStore();
+    const capturedOptions: any[] = [];
+
+    mockedCreateHaiAgent.mockImplementation(async (opts: any) => {
+      capturedOptions.push(opts);
+      return {
+        session: {
+          prompt: vi.fn().mockResolvedValue(undefined),
+          dispose: vi.fn(),
+          state: {},
+        },
+      } as any;
+    });
+
+    const executor = new TaskExecutor(store, "/tmp/test");
+
+    // Override getTask to return task with model overrides
+    store.getTask.mockResolvedValue({
+      id: "KB-001",
+      title: "Test",
+      description: "Test task",
+      column: "in-progress",
+      dependencies: [],
+      steps: [],
+      currentStep: 0,
+      log: [],
+      prompt: "# test",
+      createdAt: new Date().toISOString(),
+      updatedAt: new Date().toISOString(),
+      modelProvider: "anthropic",
+      modelId: "claude-sonnet-4-5",
+    });
+
+    await executor.execute({
+      id: "KB-001",
+      title: "Test",
+      description: "Test task",
+      column: "in-progress",
+      dependencies: [],
+      steps: [],
+      currentStep: 0,
+      log: [],
+      createdAt: new Date().toISOString(),
+      updatedAt: new Date().toISOString(),
+      modelProvider: "anthropic",
+      modelId: "claude-sonnet-4-5",
+    });
+
+    // Should use per-task model overrides
+    expect(capturedOptions[0].defaultProvider).toBe("anthropic");
+    expect(capturedOptions[0].defaultModelId).toBe("claude-sonnet-4-5");
+  });
+
+  it("falls back to global settings when per-task model is not fully specified", async () => {
+    const store = createMockStore();
+    const capturedOptions: any[] = [];
+
+    mockedCreateHaiAgent.mockImplementation(async (opts: any) => {
+      capturedOptions.push(opts);
+      return {
+        session: {
+          prompt: vi.fn().mockResolvedValue(undefined),
+          dispose: vi.fn(),
+          state: {},
+        },
+      } as any;
+    });
+
+    store.getSettings.mockResolvedValue({
+      maxConcurrent: 2,
+      maxWorktrees: 4,
+      pollIntervalMs: 15000,
+      groupOverlappingFiles: false,
+      autoMerge: false,
+      worktreeInitCommand: undefined,
+      defaultProvider: "openai",
+      defaultModelId: "gpt-4o",
+    });
+
+    const executor = new TaskExecutor(store, "/tmp/test");
+
+    await executor.execute({
+      id: "KB-001",
+      title: "Test",
+      description: "Test task",
+      column: "in-progress",
+      dependencies: [],
+      steps: [],
+      currentStep: 0,
+      log: [],
+      createdAt: new Date().toISOString(),
+      updatedAt: new Date().toISOString(),
+      // No modelProvider/modelId set
+    });
+
+    // Should use global settings (not task overrides)
+    expect(capturedOptions[0].defaultProvider).toBe("openai");
+    expect(capturedOptions[0].defaultModelId).toBe("gpt-4o");
+  });
+
+  it("falls back to global settings when only modelProvider is set (missing modelId)", async () => {
+    const store = createMockStore();
+    const capturedOptions: any[] = [];
+
+    mockedCreateHaiAgent.mockImplementation(async (opts: any) => {
+      capturedOptions.push(opts);
+      return {
+        session: {
+          prompt: vi.fn().mockResolvedValue(undefined),
+          dispose: vi.fn(),
+          state: {},
+        },
+      } as any;
+    });
+
+    store.getSettings.mockResolvedValue({
+      maxConcurrent: 2,
+      maxWorktrees: 4,
+      pollIntervalMs: 15000,
+      groupOverlappingFiles: false,
+      autoMerge: false,
+      worktreeInitCommand: undefined,
+      defaultProvider: "openai",
+      defaultModelId: "gpt-4o",
+    });
+
+    const executor = new TaskExecutor(store, "/tmp/test");
+
+    // Override getTask to return task with only modelProvider set
+    store.getTask.mockResolvedValue({
+      id: "KB-001",
+      title: "Test",
+      description: "Test task",
+      column: "in-progress",
+      dependencies: [],
+      steps: [],
+      currentStep: 0,
+      log: [],
+      prompt: "# test",
+      createdAt: new Date().toISOString(),
+      updatedAt: new Date().toISOString(),
+      modelProvider: "anthropic",
+      // modelId is missing
+    });
+
+    await executor.execute({
+      id: "KB-001",
+      title: "Test",
+      description: "Test task",
+      column: "in-progress",
+      dependencies: [],
+      steps: [],
+      currentStep: 0,
+      log: [],
+      createdAt: new Date().toISOString(),
+      updatedAt: new Date().toISOString(),
+      modelProvider: "anthropic",
+      // modelId is missing
+    });
+
+    // Should fall back to global settings since modelId is not set
+    expect(capturedOptions[0].defaultProvider).toBe("openai");
+    expect(capturedOptions[0].defaultModelId).toBe("gpt-4o");
+  });
+});
--- a/packages/engine/src/executor.ts
+++ b/packages/engine/src/executor.ts
@@ -396,7 +396,7 @@ export class TaskExecutor {
        this.createTaskCreateTool(),
        this.createTaskAddDepTool(task.id),
        this.createTaskDoneTool(task.id, () => { taskDone = true; }),
-        this.createReviewStepTool(task.id, worktreePath, detail.prompt, codeReviewVerdicts, sessionRef, stepCheckpoints),
+        this.createReviewStepTool(task.id, worktreePath, detail.prompt, codeReviewVerdicts, sessionRef, stepCheckpoints, detail),
      ];

      const agentLogger = new AgentLogger({
@@ -408,6 +408,15 @@ export class TaskExecutor {
      });

      const agentWork = async () => {
+        // Resolve model settings: use per-task overrides if both provider and modelId are set,
+        // otherwise fall back to global settings
+        const executorProvider = detail.modelProvider && detail.modelId
+          ? detail.modelProvider
+          : settings.defaultProvider;
+        const executorModelId = detail.modelProvider && detail.modelId
+          ? detail.modelId
+          : settings.defaultModelId;
+
        const { session } = await createKbAgent({
          cwd: worktreePath,
          systemPrompt: EXECUTOR_SYSTEM_PROMPT,
@@ -417,8 +426,8 @@ export class TaskExecutor {
          onThinking: agentLogger.onThinking,
          onToolStart: agentLogger.onToolStart,
          onToolEnd: agentLogger.onToolEnd,
-          defaultProvider: settings.defaultProvider,
-          defaultModelId: settings.defaultModelId,
+          defaultProvider: executorProvider,
+          defaultModelId: executorModelId,
          defaultThinkingLevel: settings.defaultThinkingLevel,
        });

@@ -732,6 +741,7 @@ export class TaskExecutor {
    codeReviewVerdicts: Map<number, ReviewVerdict>,
    sessionRef: { current: AgentSession | null },
    stepCheckpoints: Map<number, string>,
+    detail: TaskDetail,
  ): ToolDefinition {
    const store = this.store;
    const options = this.options;
@@ -761,6 +771,8 @@ export class TaskExecutor {
              defaultProvider: settings.defaultProvider,
              defaultModelId: settings.defaultModelId,
              defaultThinkingLevel: settings.defaultThinkingLevel,
+              validatorModelProvider: detail.validatorModelProvider,
+              validatorModelId: detail.validatorModelId,
              store,
              taskId,
            },
--- a/packages/engine/src/reviewer.test.ts
+++ b/packages/engine/src/reviewer.test.ts
@@ -245,3 +245,95 @@ describe("reviewStep — exhausted-retry error detection", () => {
    expect(result.verdict).toBe("APPROVE");
  });
 });
+
+describe("reviewStep — validator model overrides", () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+  });
+
+  it("uses validatorModelProvider and validatorModelId when both are set", async () => {
+    mockedCreateHaiAgent.mockResolvedValue(
+      createMockSession("### Verdict: APPROVE\n### Summary\nLooks good."),
+    );
+
+    await reviewStep(
+      "/tmp/worktree", "KB-100", 1, "Test Step", "plan", "# prompt",
+      undefined,
+      {
+        defaultProvider: "openai",
+        defaultModelId: "gpt-4o",
+        validatorModelProvider: "anthropic",
+        validatorModelId: "claude-sonnet-4-5",
+      },
+    );
+
+    expect(mockedCreateHaiAgent).toHaveBeenCalledTimes(1);
+    const opts = mockedCreateHaiAgent.mock.calls[0][0];
+    expect(opts.defaultProvider).toBe("anthropic");
+    expect(opts.defaultModelId).toBe("claude-sonnet-4-5");
+  });
+
+  it("falls back to defaultProvider/defaultModelId when validatorModelProvider is missing", async () => {
+    mockedCreateHaiAgent.mockResolvedValue(
+      createMockSession("### Verdict: APPROVE\n### Summary\nLooks good."),
+    );
+
+    await reviewStep(
+      "/tmp/worktree", "KB-100", 1, "Test Step", "plan", "# prompt",
+      undefined,
+      {
+        defaultProvider: "openai",
+        defaultModelId: "gpt-4o",
+        // validatorModelProvider is missing
+        validatorModelId: "claude-sonnet-4-5",
+      },
+    );
+
+    expect(mockedCreateHaiAgent).toHaveBeenCalledTimes(1);
+    const opts = mockedCreateHaiAgent.mock.calls[0][0];
+    expect(opts.defaultProvider).toBe("openai");
+    expect(opts.defaultModelId).toBe("gpt-4o");
+  });
+
+  it("falls back to defaultProvider/defaultModelId when validatorModelId is missing", async () => {
+    mockedCreateHaiAgent.mockResolvedValue(
+      createMockSession("### Verdict: APPROVE\n### Summary\nLooks good."),
+    );
+
+    await reviewStep(
+      "/tmp/worktree", "KB-100", 1, "Test Step", "plan", "# prompt",
+      undefined,
+      {
+        defaultProvider: "openai",
+        defaultModelId: "gpt-4o",
+        validatorModelProvider: "anthropic",
+        // validatorModelId is missing
+      },
+    );
+
+    expect(mockedCreateHaiAgent).toHaveBeenCalledTimes(1);
+    const opts = mockedCreateHaiAgent.mock.calls[0][0];
+    expect(opts.defaultProvider).toBe("openai");
+    expect(opts.defaultModelId).toBe("gpt-4o");
+  });
+
+  it("falls back to defaultProvider/defaultModelId when both validator fields are undefined", async () => {
+    mockedCreateHaiAgent.mockResolvedValue(
+      createMockSession("### Verdict: APPROVE\n### Summary\nLooks good."),
+    );
+
+    await reviewStep(
+      "/tmp/worktree", "KB-100", 1, "Test Step", "plan", "# prompt",
+      undefined,
+      {
+        defaultProvider: "openai",
+        defaultModelId: "gpt-4o",
+      },
+    );
+
+    expect(mockedCreateHaiAgent).toHaveBeenCalledTimes(1);
+    const opts = mockedCreateHaiAgent.mock.calls[0][0];
+    expect(opts.defaultProvider).toBe("openai");
+    expect(opts.defaultModelId).toBe("gpt-4o");
+  });
+});
--- a/packages/engine/src/reviewer.ts
+++ b/packages/engine/src/reviewer.ts
@@ -144,6 +144,10 @@ export interface ReviewOptions {
  defaultProvider?: string;
  /** Default model ID within the provider (e.g. "claude-sonnet-4-5"). When set with `defaultProvider`, overrides the reviewer's model selection. */
  defaultModelId?: string;
+  /** Validator model provider override. When both `validatorModelProvider` and `validatorModelId` are set, they take precedence over `defaultProvider`/`defaultModelId`. */
+  validatorModelProvider?: string;
+  /** Validator model ID override. When both `validatorModelProvider` and `validatorModelId` are set, they take precedence over `defaultProvider`/`defaultModelId`. */
+  validatorModelId?: string;
  /** Default thinking effort level for the reviewer agent session. */
  defaultThinkingLevel?: string;
  /** Task store for persisting agent log entries. When provided with `taskId`, enables full conversation logging. */
@@ -182,6 +186,15 @@ export async function reviewStep(
      })
    : null;

+  // Resolve validator model settings: use per-task overrides if both provider and modelId are set,
+  // otherwise fall back to defaultProvider/defaultModelId
+  const validatorProvider = options.validatorModelProvider && options.validatorModelId
+    ? options.validatorModelProvider
+    : options.defaultProvider;
+  const validatorModelId = options.validatorModelProvider && options.validatorModelId
+    ? options.validatorModelId
+    : options.defaultModelId;
+
  // Spawn a reviewer agent with read-only tools
  const { session } = await createKbAgent({
    cwd,
@@ -191,8 +204,8 @@ export async function reviewStep(
    onThinking: agentLogger?.onThinking,
    onToolStart: agentLogger?.onToolStart,
    onToolEnd: agentLogger?.onToolEnd,
-    defaultProvider: options.defaultProvider,
-    defaultModelId: options.defaultModelId,
+    defaultProvider: validatorProvider,
+    defaultModelId: validatorModelId,
    defaultThinkingLevel: options.defaultThinkingLevel,
  });