FN-6235: source reviewer prompts from workflow IR

Deduplicate reviewer policy text by making workflow IR review seams the engine's built-in prompt source.

- Move the canonical built-in reviewer prompt into core agent prompts and export seam prompt resolution helpers.
- Resolve reviewer prompts from explicit role overrides first, then workflow IR review seams, with the built-in prompt as a fallback.
- Update prompt cache and reviewer tests to cover single-source reviewer prompt behavior.
- Add a patch changeset for the published Fusion package.

Files changed:
 .../FN-6235-reviewer-prompt-single-source.md       |   5 +
 packages/core/src/__tests__/agent-prompts.test.ts  |   8 +-
 packages/core/src/agent-prompts.ts                 | 109 +++++++++--
 packages/core/src/index.ts                         |   2 +
 packages/core/src/workflow-ir-resolver.ts          |  31 ++-
 .../src/__tests__/prompt-cache-integration.test.ts |  10 +-
 .../reviewer-prompt-single-source.test.ts          | 160 +++++++++++++++
 packages/engine/src/__tests__/reviewer.test.ts     |  97 ++++-----
 packages/engine/src/prompt-layers.ts               |   2 +-
 packages/engine/src/reviewer.ts                    | 217 ++-------------------
 10 files changed, 370 insertions(+), 271 deletions(-)

Fusion-Task-Id: FN-6235

Fusion-Task-Lineage: 725932d1-2469-4507-a0f9-0946c83a1572

.changeset/FN-6235-reviewer-prompt-single-source.md

@@ -0,0 +1,5 @@
+---
+"@runfusion/fusion": patch
+---
+
+Resolve the built-in reviewer base prompt from the workflow IR `review` node instead of an engine-local `REVIEWER_SYSTEM_PROMPT` duplicate. The canonical reviewer policy now lives in the `default-reviewer` agent prompt / built-in workflow seam, with reconciled superset content that preserves the FN-5928/FN-6229 surface-enumeration and symptom-verification gates, undersplit-task guidance, test-quality rules, worktree-boundary review, and the embedded port-4040 safety rule.

packages/core/src/__tests__/agent-prompts.test.ts

@@ -10,7 +10,7 @@ import {
 } from "../agent-prompts.js";
 import { BUILTIN_CODING_WORKFLOW_IR } from "../builtin-coding-workflow-ir.js";
 import { renderTriagePolicyPlaceholders } from "../builtin-workflow-settings.js";
-import { resolvePlanningPromptFromIr } from "../workflow-ir-resolver.js";
+import { resolvePlanningPromptFromIr, resolveSeamPromptFromIr } from "../workflow-ir-resolver.js";
 import type { AgentPromptsConfig, AgentPromptTemplate } from "../types.js";
 import type { WorkflowIr } from "../workflow-ir-types.js";
 
@@ -286,13 +286,14 @@ describe("resolveAgentPrompt", () => {
     expect(renderedPrompt).not.toContain("{{");
   });
 
-  it("resolves custom planning prompts and ignores IRs without planning prompts", () => {
+  it("resolves custom seam prompts and ignores IRs without matching prompts", () => {
     const customIr: WorkflowIr = {
       version: "v1",
       name: "custom",
       nodes: [
         { id: "start", kind: "start" },
         { id: "planning", kind: "prompt", config: { seam: "planning", prompt: "custom planning prompt" } },
+        { id: "review", kind: "prompt", config: { seam: "review", prompt: "custom review prompt" } },
       ],
       edges: [],
     };
@@ -304,7 +305,10 @@ describe("resolveAgentPrompt", () => {
     };
 
     expect(resolvePlanningPromptFromIr(customIr)).toBe("custom planning prompt");
+    expect(resolveSeamPromptFromIr(customIr, "review")).toBe("custom review prompt");
+    expect(resolveSeamPromptFromIr(BUILTIN_CODING_WORKFLOW_IR, "review")).toBe(resolveAgentPrompt("reviewer"));
     expect(resolvePlanningPromptFromIr(noPlanningIr)).toBeUndefined();
+    expect(resolveSeamPromptFromIr(noPlanningIr, "review")).toBeUndefined();
   });
 
   it("built-in triage prompt requires surface enumeration for bug-fix specs", () => {

packages/core/src/agent-prompts.ts

@@ -7,11 +7,10 @@
  * - Additional role variants (senior-engineer, strict-reviewer, concise-triage)
  * - A resolver function that merges custom templates from project settings with built-ins
  *
- * NOTE: The built-in prompt texts are derived from the engine's hardcoded prompts
+ * NOTE: Built-in prompt texts that feed workflow seams live here as the canonical
- * (EXECUTOR_SYSTEM_PROMPT, TRIAGE_SYSTEM_PROMPT, REVIEWER_SYSTEM_PROMPT, and the
+ * source for @fusion/core and @fusion/engine. Engine code should resolve triage
- * merger prompt). They should be kept in sync when the engine prompts change.
+ * and reviewer built-ins through workflow IR seam prompts instead of carrying
- * Since @fusion/core cannot import @fusion/engine (circular dependency), these
+ * duplicate policy constants.
- * are maintained as inline strings.
  *
  * @module agent-prompts
  */
@@ -19,7 +18,7 @@
 import type { AgentCapability, AgentPromptTemplate, AgentPromptsConfig } from "./types.js";
 
 // ---------------------------------------------------------------------------
-// Built-in prompt text (derived from engine constants — keep in sync)
+// Built-in prompt text (canonical source for workflow seam prompts)
 // ---------------------------------------------------------------------------
 
 const EXECUTOR_PROMPT_TEXT = `You are a task execution agent for "fn", an AI-orchestrated task board.
@@ -538,11 +537,26 @@ Use this exact checklist (keep it verbatim — do not expand or reorder):
 
 Only inject this section when the task genuinely touches frontend UI. Omit it for backend-only, config-only, or documentation-only tasks.`;;
 
+// FN-6235: single source for the built-in reviewer policy; the engine REVIEWER_SYSTEM_PROMPT duplicate was removed.
 const REVIEWER_PROMPT_TEXT = `You are an independent code and plan reviewer.
 
+## Your Role
+You are an objective quality gate for plans, code, and specs.
+You are neither the implementor's advocate nor adversary: your job is evidence-based assessment that protects delivery quality.
+
 You provide quality assessment for task implementations. You have full read
 access to the codebase and can run commands to inspect code.
 
+## What to Look For
+- Correctness against stated requirements
+- Edge-case handling and failure-path behavior
+- Test adequacy (behavior-focused coverage, meaningful assertions)
+- Consistency with existing project patterns and conventions
+- Security, data-safety, and permission boundary concerns
+- Performance implications where changes affect hot paths or heavy operations
+
+Review efficiently: prioritize high-impact correctness/risk issues first. Do not spend blocking attention on style nits when substantive defects exist.
+
 ## Verdict Criteria
 
 - **APPROVE** — Step will achieve its stated outcomes. Minor suggestions go in
@@ -556,6 +570,11 @@ access to the codebase and can run commands to inspect code.
 
 ### APPROVE vs REVISE
 
+Concrete examples:
+- APPROVE: implementation satisfies outcomes; only optional cleanup or minor wording suggestions remain.
+- REVISE: a required behavior is missing, tests are insufficient for changed behavior, or a likely regression exists.
+- RETHINK: the approach conflicts with architecture/task goals such that incremental edits are unlikely to rescue it.
+
 **APPROVE** when:
 - The approach will work, but you see a cleaner alternative
 - Documentation style could improve
@@ -568,6 +587,7 @@ access to the codebase and can run commands to inspect code.
 - Backward compatibility is broken without migration
 - Code outside the task's File Scope is deleted, removed, or gutted (out-of-scope removal)
 - Existing functionality is removed without a corresponding changeset explaining the removal
+- Code changes were made outside the assigned task worktree, unless the path is an expected exception such as project memory or task attachments
 
 ### Do NOT issue REVISE for
 - STATUS/formatting preferences
@@ -610,7 +630,7 @@ access to the codebase and can run commands to inspect code.
 
 ### Test Gaps
 - [Missing test scenarios]
-- [For bug fixes, call out any repro-only regression test that does not assert the invariant across the enumerated surfaces. Issue REVISE when coverage stops at the single reported case instead of spanning the \`## Surface Enumeration\` checklist (FN-5893; see FN-5787/FN-5789/FN-5803, FN-5797/FN-5875/FN-5919, and FN-5751).]
+- [For bug fixes and UI-affordance add/remove changes, call out any single-surface-only test that doesn't verify the invariant across the spec's enumerated surfaces. For UI-affordance removals, also flag tests that don't verify the removed affordance's container/wrapper is fully cleaned up on both desktop and mobile breakpoints. Issue REVISE when coverage stops at the single reported surface (FN-6134; see FN-6115→FN-6118→FN-6123 for the motivating multi-task incident). Keep enforcing FN-5893 for bug fixes; see FN-5787/FN-5789/FN-5803, FN-5797/FN-5875/FN-5919, and FN-5751.]
 
 ### Suggestions
 - [Optional improvements, not blocking]
@@ -635,18 +655,85 @@ access to the codebase and can run commands to inspect code.
 - **File scope accuracy:** [All affected files listed? No extras?]
 - **Dependency correctness:** [Dependencies exist and are appropriate?]
 - **Testing requirements:** [Real automated tests required, not just typechecks?]
-- **Surface enumeration:** [For bug-fix specs, is \`## Surface Enumeration\` present and does it enumerate the relevant providers/bridges/execution paths, desktop + mobile breakpoints/platforms, empty/undefined/duplicate/populated states, and shared hooks/components/modules/helpers? Missing or incomplete coverage is a blocking REVISE.]
+- **Surface enumeration:** [For bug-fix specs and UI-affordance add/remove specs, is \`## Surface Enumeration\` present and does it enumerate the relevant providers/bridges/execution paths, desktop + mobile breakpoints/platforms, empty/undefined/duplicate/populated states, and shared hooks/components/modules/helpers? For UI-affordance add/remove tasks, also verify: (a) the spec searches for ALL components rendering the affordance, not just the one the user pointed at; (b) the spec explicitly addresses leftover shells after removal across desktop and mobile breakpoints. Missing or incomplete coverage is a blocking REVISE.]
+- **Symptom verification:** [For bug-class/bug-fix specs only, is \`## Symptom Verification\` present and complete with **Original symptom**, **Exact reproduction**, and **Assertion it is gone**? A bug-class spec whose final verification only checks green build/tests without reproducing the original failure and asserting it no longer occurs is a blocking REVISE under FN-5893. Missing, empty, or incomplete \`## Symptom Verification\` is a blocking REVISE for bug-class specs; feature/docs/non-bug specs are not required to carry it.]
 - **Documentation completeness:** [Must Update / Check If Affected sections present?]
+- **Dangling task-document references:** [No \`.fusion/tasks/<id>/<file>\` path is cited in Context, Steps, or File Scope unless the file exists or is explicitly created as a \`(new)\` artifact in this spec. References to nonexistent task-local artifacts are a blocking REVISE.]
 - **Sizing & review level:** [Size and review level appropriate for the work?]
-- **Subtask breakdown:** [Were complex tasks appropriately split into 2-5 child tasks? A task with 8+ implementation steps, affecting 3+ packages, should have been divided]
+- **Subtask breakdown:** [Only flag genuinely oversized specs (12+ implementation steps, OR 5+ truly independent deliverables that could ship separately). Do NOT flag a coherent vertical change just because it touches multiple packages. When borderline, prefer leaving the task whole.]
 - **User comment coverage:** [Were all user comments addressed? Every user comment must be reflected in the spec — missing coverage is a blocking REVISE]
 
 ### Suggestions
 - [Optional improvements, not blocking]
 \`\`\`
 
-## Safety Rules
+## Spec Review — Undersplit Task Detection
-- **NEVER kill processes on port 4040.** Port 4040 is the production dashboard. If you need to test server endpoints, start a server on a different port (\`--port 0\` for random). If port 4040 is occupied, use a different port — do NOT kill the occupant. Issue REVISE if the executor kills or attempts to kill processes on port 4040.`;
+
+When reviewing specs, assess whether the task should have been broken into subtasks. The bar for splitting is high — most tasks should remain whole. Coordination overhead (worktrees, dependency wiring, merge sequencing) is real, so splitting must clearly pay for itself.
+
+**Default position:** do NOT flag undersplit. Reach for it only when the spec is genuinely oversized.
+
+**Flag as REVISE only when ALL of the following are true:**
+- The spec has 12+ implementation steps, OR contains 5+ clearly independent deliverables that could be shipped separately by different people
+- The deliverables are NOT a coherent vertical change (a single feature touching core + dashboard + tests is coherent — do not split it)
+- Splitting would produce children that each have ≥4 steps and a clearly distinct scope
+
+If the spec is borderline (under those thresholds, or arguable), put your splitting suggestion in the **Suggestions** section instead of REVISE — the planner can take it or leave it.
+
+**How to flag an undersplit task (only when the criteria above are met):**
+Say explicitly: "This task should be broken into subtasks because [specific reason]."
+Recommend the number of child tasks (2-5) and what each should cover.
+Instruct the planner to:
+1. Use the \`fn_task_create\` tool to create 2–5 child tasks from the oversized spec
+2. Do NOT write a parent PROMPT.md — the parent will be closed automatically after children are created
+   (Not write a parent PROMPT.md is also unacceptable.)
+3. Make each child cover one coherent deliverable with clear scope boundaries
+
+Example REVISE feedback for a genuinely oversized task:
+"This task has 14 steps and contains 4 independent deliverables (engine integration, dashboard UI, CLI command, migration tooling) that could ship separately. Use fn_task_create to split into: (1) engine logic, (2) dashboard UI, (3) CLI integration, (4) migration tooling. Do not write a parent PROMPT."
+
+**Do NOT flag if ANY of these apply:**
+- The spec has 11 or fewer implementation steps
+- Steps are sequential and tightly coupled (e.g., a pipeline where each step depends on the previous)
+- The task is a vertical change touching multiple packages for one coherent feature (typical in this monorepo)
+- The task is a bug fix, regardless of how many files it touches
+- Splitting would create coordination overhead that exceeds the benefit
+
+## Plan Granularity
+
+When reviewing plans, assess whether the approach achieves the step's OUTCOMES —
+not whether every function and parameter is listed.
+
+Good plan: identifies key behavioral changes, calls out risks, has a testing strategy.
+Do NOT demand function-level implementation checklists.
+
+## Test Quality Review
+
+When reviewing tests, check that they verify observable behavior and regression risk (not only implementation trivia).
+Flag REVISE when key edge cases or failure modes for changed behavior are untested.
+For bug fixes, apply FN-5893 strictly: if the regression test only reproduces the reported case instead of asserting the invariant across the spec's \`## Surface Enumeration\` surfaces, issue REVISE. Treat that as a repro-only regression test; issue REVISE when coverage stops at the single reported case instead of spanning the \`## Surface Enumeration\` checklist. Use the motivating recurrences (FN-5787/FN-5789/FN-5803, FN-5797/FN-5875/FN-5919, and FN-5751) as concrete examples of why repro-only coverage is insufficient.
+For bug-class/bug-fix specs, also enforce symptom-based acceptance: if the spec is missing \`## Symptom Verification\`, leaves it empty/incomplete, lacks **Original symptom**, **Exact reproduction**, or **Assertion it is gone**, or its final verification only checks green build/tests without reproducing the original failure condition and asserting it no longer occurs, issue REVISE. Do not require \`## Symptom Verification\` for feature/docs/non-bug specs.
+For UI-affordance add/remove changes, apply the same surface-enumeration strictness: if the test only checks the single surface the user reported instead of all enumerated surfaces, issue REVISE. For UI-affordance removals, require coverage/evidence that empty button shells, orphaned click targets, now-unused wrappers, and dangling aria-labels are cleaned up across desktop and mobile breakpoints; FN-6115/FN-6118/FN-6123 is the motivating recurrence.
+
+## Worktree Boundary Review
+
+For code reviews, verify that implementation changes are in the assigned task
+worktree. The review request includes the current worktree path. Inspect git
+state and recent commits from that worktree, and treat changes outside it as a
+blocking REVISE unless they are expected project-root state such as
+\`.fusion/memory/\` files, task attachments, or other explicitly documented
+Fusion metadata. If you see edits or commits in the primary project checkout
+instead of the task worktree, call that out directly and ask the worker to move
+the changes into the assigned worktree.
+
+## Rules
+
+- Be specific — reference actual files and line numbers
+- Be constructive — suggest fixes, not just problems
+- Be proportional — don't block on style nits
+- Output your review as plain text (not to a file)
+- **NEVER kill processes on port 4040.** Port 4040 is the production dashboard. If you need to test server endpoints, start a server on a different port (\`--port 0\` for random). If port 4040 is occupied, use a different port — do NOT kill the occupant. Issue REVISE if the executor kills or attempts to kill processes on port 4040.
+`;
 
 /**
  * Base merger prompt text (without commit format instructions, which are

packages/core/src/index.ts

@@ -314,7 +314,9 @@ export {
 export {
   resolveWorkflowIrForTask,
   resolveWorkflowIrById,
+  resolveSeamPromptFromIr,
   resolvePlanningPromptFromIr,
+  resolveTaskSeamPrompt,
   resolveTaskPlanningPrompt,
   type WorkflowIrResolverStore,
 } from "./workflow-ir-resolver.js";

packages/core/src/workflow-ir-resolver.ts

@@ -26,31 +26,50 @@ export interface WorkflowIrResolverStore {
 }
 
 /**
- * Extract the planning seam prompt from a resolved workflow IR.
+ * Extract a prompt seam's prompt text from a resolved workflow IR.
  *
- * Planning seam nodes are prompt nodes with `config.seam === "planning"`;
+ * Seam prompt nodes are prompt nodes with `config.seam === seam`;
  * `config.prompt` carries the text installed by builtinPromptConfig or a custom
  * workflow author. Empty/missing prompts return undefined so callers can apply
  * their own fail-soft fallback.
  */
-export function resolvePlanningPromptFromIr(ir: WorkflowIr): string | undefined {
+export function resolveSeamPromptFromIr(ir: WorkflowIr, seam: string): string | undefined {
   for (const node of ir.nodes) {
     if (node.kind !== "prompt") continue;
-    if (node.config?.seam !== "planning") continue;
+    if (node.config?.seam !== seam) continue;
     const prompt = node.config.prompt;
     if (typeof prompt === "string" && prompt.trim().length > 0) return prompt;
   }
   return undefined;
 }
 
+/** Extract the planning seam prompt from a resolved workflow IR. */
+export function resolvePlanningPromptFromIr(ir: WorkflowIr): string | undefined {
+  return resolveSeamPromptFromIr(ir, "planning");
+}
+
+/** Resolve a task's seam prompt via its selected workflow IR. */
+export async function resolveTaskSeamPrompt(
+  store: WorkflowIrResolverStore,
+  taskId: string,
+  seam: string,
+  irCache?: Map<string, WorkflowIr>,
+): Promise<string | undefined> {
+  try {
+    const ir = await resolveWorkflowIrForTask(store, taskId, irCache);
+    return resolveSeamPromptFromIr(ir, seam);
+  } catch {
+    return undefined;
+  }
+}
+
 /** Resolve a task's planning seam prompt via its selected workflow IR. */
 export async function resolveTaskPlanningPrompt(
   store: WorkflowIrResolverStore,
   taskId: string,
   irCache?: Map<string, WorkflowIr>,
 ): Promise<string | undefined> {
-  const ir = await resolveWorkflowIrForTask(store, taskId, irCache);
+  return resolveTaskSeamPrompt(store, taskId, "planning", irCache);
-  return resolvePlanningPromptFromIr(ir);
 }
 
 /**

packages/engine/src/__tests__/prompt-cache-integration.test.ts

@@ -1,13 +1,15 @@
 import { describe, it, expect } from "vitest";
+import { resolveAgentPrompt } from "@fusion/core";
 import { buildPromptLayers, collapsePromptLayers, type SystemPromptLayers } from "../prompt-layers.js";
-import { REVIEWER_SYSTEM_PROMPT } from "../reviewer.js";
+
+const DEFAULT_REVIEWER_PROMPT = resolveAgentPrompt("reviewer");
 
 describe("cross-session prompt cache integration", () => {
   const MEMORY_INSTRUCTIONS = "\n## Memory\n\nUse fn_memory_search to look up relevant context.";
 
   function simulateReviewerSession(sessionIndex: number): SystemPromptLayers {
     return buildPromptLayers({
-      basePrompt: REVIEWER_SYSTEM_PROMPT,
+      basePrompt: DEFAULT_REVIEWER_PROMPT,
       agentInstructions: `Session ${sessionIndex}: custom instructions that vary per agent.`,
       memorySection: MEMORY_INSTRUCTIONS,
       pluginContributions: sessionIndex % 2 === 0
@@ -42,8 +44,8 @@ describe("cross-session prompt cache integration", () => {
     }
   });
 
-  it("stable prefix starts with REVIEWER_SYSTEM_PROMPT", () => {
+  it("stable prefix starts with the canonical default reviewer prompt", () => {
     const layers = simulateReviewerSession(0);
-    expect(layers.stable.startsWith(REVIEWER_SYSTEM_PROMPT)).toBe(true);
+    expect(layers.stable.startsWith(DEFAULT_REVIEWER_PROMPT)).toBe(true);
   });
 });

packages/engine/src/__tests__/reviewer-prompt-single-source.test.ts

@@ -0,0 +1,160 @@
+import { readFileSync } from "node:fs";
+import { resolve } from "node:path";
+import { fileURLToPath } from "node:url";
+import { describe, it, expect, vi, beforeEach } from "vitest";
+import {
+  BUILTIN_CODING_WORKFLOW_IR,
+  resolveAgentPrompt,
+  resolveSeamPromptFromIr,
+  type WorkflowIr,
+} from "@fusion/core";
+
+vi.mock("../pi.js", () => ({
+  createFnAgent: vi.fn(),
+  describeModel: vi.fn().mockReturnValue("mock-provider/mock-model"),
+  promptWithFallback: vi.fn(async (session, prompt, options) => {
+    if (options === undefined) {
+      await session.prompt(prompt);
+    } else {
+      await session.prompt(prompt, options);
+    }
+  }),
+}));
+
+import { reviewStep } from "../reviewer.js";
+import { createFnAgent } from "../pi.js";
+
+const mockedCreateFnAgent = vi.mocked(createFnAgent);
+
+function createMockSession(reviewText = "### Verdict: APPROVE\n### Summary\nLooks good.") {
+  return {
+    session: {
+      prompt: vi.fn().mockResolvedValue(undefined),
+      subscribe: vi.fn().mockImplementation((cb: any) => {
+        cb({
+          type: "message_update",
+          assistantMessageEvent: { type: "text_delta", delta: reviewText },
+        });
+      }),
+      dispose: vi.fn(),
+    },
+  } as any;
+}
+
+function createStore(workflowId = "builtin:coding", customIr?: WorkflowIr) {
+  return {
+    getSettings: vi.fn().mockResolvedValue({}),
+    getTaskWorkflowSelection: vi.fn().mockReturnValue({ workflowId, stepIds: [] }),
+    getWorkflowDefinition: vi.fn().mockImplementation(async (id: string) => {
+      if (customIr && id === workflowId) return { ir: customIr };
+      return undefined;
+    }),
+  } as any;
+}
+
+async function captureReviewerSystemPrompt(options: Parameters<typeof reviewStep>[7] = {}) {
+  mockedCreateFnAgent.mockResolvedValue(createMockSession());
+  await reviewStep(
+    "/tmp/worktree",
+    "FN-6235",
+    1,
+    "Review prompt source",
+    "plan",
+    "# Plan",
+    undefined,
+    options,
+  );
+  return mockedCreateFnAgent.mock.calls[0][0].systemPrompt as string;
+}
+
+beforeEach(() => {
+  vi.clearAllMocks();
+});
+
+describe("reviewer prompt single source", () => {
+  it("does not reintroduce an engine reviewer policy constant", () => {
+    const reviewerSource = readFileSync(
+      resolve(fileURLToPath(new URL("..", import.meta.url)), "reviewer.ts"),
+      "utf8",
+    );
+
+    expect(reviewerSource).not.toMatch(/export const REVIEWER_SYSTEM_PROMPT\s*=/);
+    expect(reviewerSource).not.toMatch(/export const [A-Z_]*REVIEWER[A-Z_]*SYSTEM_PROMPT\s*=/);
+  });
+
+  it("keeps builtin coding review seam byte-identical to the default reviewer prompt", () => {
+    expect(resolveSeamPromptFromIr(BUILTIN_CODING_WORKFLOW_IR, "review")).toBe(resolveAgentPrompt("reviewer"));
+  });
+
+  it("uses the builtin coding IR review-node prompt when no user override is set", async () => {
+    const systemPrompt = await captureReviewerSystemPrompt({ store: createStore() });
+
+    expect(systemPrompt).toBe(resolveSeamPromptFromIr(BUILTIN_CODING_WORKFLOW_IR, "review"));
+  });
+
+  it("uses a selected custom workflow review-node prompt", async () => {
+    const customIr: WorkflowIr = {
+      version: "v1",
+      name: "custom-reviewer",
+      nodes: [
+        { id: "start", kind: "start" },
+        { id: "review", kind: "prompt", config: { seam: "review", prompt: "custom workflow reviewer prompt" } },
+      ],
+      edges: [],
+    };
+
+    const systemPrompt = await captureReviewerSystemPrompt({ store: createStore("WF-review", customIr) });
+
+    expect(systemPrompt).toBe("custom workflow reviewer prompt");
+  });
+
+  it("preserves reviewer user-override precedence over workflow IR prompts", async () => {
+    const customIr: WorkflowIr = {
+      version: "v1",
+      name: "custom-reviewer",
+      nodes: [
+        { id: "review", kind: "prompt", config: { seam: "review", prompt: "workflow prompt should not win" } },
+      ],
+      edges: [],
+    };
+
+    const systemPrompt = await captureReviewerSystemPrompt({
+      store: createStore("WF-review", customIr),
+      agentPrompts: {
+        templates: [{
+          id: "custom-reviewer",
+          name: "Custom Reviewer",
+          description: "Project reviewer override",
+          role: "reviewer",
+          prompt: "user override reviewer prompt",
+        }],
+        roleAssignments: { reviewer: "custom-reviewer" },
+      },
+    });
+
+    expect(systemPrompt).toBe("user override reviewer prompt");
+  });
+
+  it("falls back to a non-empty default reviewer prompt when no store is provided", async () => {
+    const systemPrompt = await captureReviewerSystemPrompt();
+
+    expect(systemPrompt).toBe(resolveAgentPrompt("reviewer"));
+    expect(systemPrompt.trim().length).toBeGreaterThan(0);
+  });
+
+  it.each(["plan", "code", "spec"] as const)("uses the same resolved base prompt for %s reviews", async (reviewType) => {
+    mockedCreateFnAgent.mockResolvedValue(createMockSession());
+    await reviewStep(
+      "/tmp/worktree",
+      "FN-6235",
+      1,
+      "Review prompt source",
+      reviewType,
+      "# Prompt",
+      undefined,
+      { store: createStore() },
+    );
+
+    expect(mockedCreateFnAgent.mock.calls[0][0].systemPrompt).toBe(resolveAgentPrompt("reviewer"));
+  });
+});

packages/engine/src/__tests__/reviewer.test.ts

@@ -12,9 +12,12 @@ vi.mock("../pi.js", () => ({
   }),
 }));
 
-import { reviewStep, REVIEWER_SYSTEM_PROMPT } from "../reviewer.js";
+import { resolveAgentPrompt } from "@fusion/core";
+import { reviewStep } from "../reviewer.js";
 import { createFnAgent, promptWithFallback } from "../pi.js";
 
+const DEFAULT_REVIEWER_PROMPT = resolveAgentPrompt("reviewer");
+
 const mockedCreateFnAgent = vi.mocked(createFnAgent);
 const mockedPromptWithFallback = vi.mocked(promptWithFallback);
 const CONTEXT_LIMIT_ERROR = "exceeded model token limit: 262144 (requested: 262879)";
@@ -293,55 +296,55 @@ describe("reviewStep — spec review type", () => {
 
 describe("FN-5928 surface-enumeration review-gate wording", () => {
   it("requires spec reviews to block missing or incomplete surface enumeration for bug-fix specs", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("**Surface enumeration:**");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("**Surface enumeration:**");
-    expect(REVIEWER_SYSTEM_PROMPT).toMatch(
+    expect(DEFAULT_REVIEWER_PROMPT).toMatch(
       /For bug-fix specs and UI-affordance add\/remove specs, is `## Surface Enumeration` present[\s\S]*Missing or incomplete coverage is a blocking REVISE\./,
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("desktop + mobile breakpoints/platforms");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("desktop + mobile breakpoints/platforms");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("shared hooks/components/modules/helpers");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("shared hooks/components/modules/helpers");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("bug-fix specs and UI-affordance add/remove specs");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("bug-fix specs and UI-affordance add/remove specs");
   });
 
   it("requires code reviews to reject repro-only regression tests for bug fixes", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toMatch(
+    expect(DEFAULT_REVIEWER_PROMPT).toMatch(
       /For bug fixes, apply FN-5893 strictly: if the regression test only reproduces the reported case instead of asserting the invariant across the spec's `## Surface Enumeration` surfaces, issue REVISE\./,
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("single-surface-only test");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("single-surface-only test");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("doesn't verify the invariant across the spec's enumerated surfaces");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("doesn't verify the invariant across the spec's enumerated surfaces");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("Keep enforcing FN-5893 for bug fixes");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("Keep enforcing FN-5893 for bug fixes");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("FN-5787/FN-5789/FN-5803");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("FN-5787/FN-5789/FN-5803");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("FN-5797/FN-5875/FN-5919");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("FN-5797/FN-5875/FN-5919");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("FN-5751");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("FN-5751");
   });
 
   it("requires spec reviews to block bug-class specs missing symptom verification", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("**Symptom verification:**");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("**Symptom verification:**");
-    expect(REVIEWER_SYSTEM_PROMPT).toMatch(
+    expect(DEFAULT_REVIEWER_PROMPT).toMatch(
       /For bug-class\/bug-fix specs only, is `## Symptom Verification` present and complete with \*\*Original symptom\*\*, \*\*Exact reproduction\*\*, and \*\*Assertion it is gone\*\*\?/,
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "A bug-class spec whose final verification only checks green build/tests without reproducing the original failure and asserting it no longer occurs is a blocking REVISE under FN-5893",
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "Missing, empty, or incomplete `## Symptom Verification` is a blocking REVISE for bug-class specs",
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("feature/docs/non-bug specs are not required to carry it");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("feature/docs/non-bug specs are not required to carry it");
   });
 
   it("requires code reviews to reject green-build-only symptom acceptance for bug fixes", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toMatch(
+    expect(DEFAULT_REVIEWER_PROMPT).toMatch(
       /For bug-class\/bug-fix specs, also enforce symptom-based acceptance:[\s\S]*final verification only checks green build\/tests without reproducing the original failure condition and asserting it no longer occurs, issue REVISE\./,
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("lacks **Original symptom**, **Exact reproduction**, or **Assertion it is gone**");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("lacks **Original symptom**, **Exact reproduction**, or **Assertion it is gone**");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("Do not require `## Symptom Verification` for feature/docs/non-bug specs");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("Do not require `## Symptom Verification` for feature/docs/non-bug specs");
   });
 
   it("requires spec/code reviews to enforce surface enumeration for UI-affordance add/remove tasks", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("leftover shells after removal");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("leftover shells after removal");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("For bug fixes and UI-affordance add/remove changes");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("For bug fixes and UI-affordance add/remove changes");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("UI-affordance removals");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("UI-affordance removals");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("For UI-affordance add/remove changes, apply the same surface-enumeration strictness");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("For UI-affordance add/remove changes, apply the same surface-enumeration strictness");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("FN-6115/FN-6118/FN-6123");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("FN-6115/FN-6118/FN-6123");
   });
 
   it("demonstrates the gate firing on a single-component UI-removal spec", () => {
@@ -349,11 +352,11 @@ describe("FN-5928 surface-enumeration review-gate wording", () => {
       "## Mission\nRemove the workflow-row chevron from WorkflowRow.tsx only.";
 
     expect(singleComponentRemovalSpec).toContain("WorkflowRow.tsx only");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("searches for ALL components rendering the affordance");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("searches for ALL components rendering the affordance");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("not just the one the user pointed at");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("not just the one the user pointed at");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("leftover shells after removal");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("leftover shells after removal");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("empty button shells");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("empty button shells");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("Issue REVISE when coverage stops at the single reported surface");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("Issue REVISE when coverage stops at the single reported surface");
   });
 });
 
@@ -876,24 +879,24 @@ describe("reviewStep — validator model overrides", () => {
   });
 });
 
-describe("REVIEWER_SYSTEM_PROMPT", () => {
+describe("default reviewer prompt", () => {
   it("includes subtask breakdown criterion in spec review", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("Subtask breakdown");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("Subtask breakdown");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "12+ implementation steps",
     );
   });
 
   it("biases the reviewer toward keeping tasks whole", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("The bar for splitting is high");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("The bar for splitting is high");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "Default position:** do NOT flag undersplit",
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("12+ implementation steps");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("12+ implementation steps");
   });
 
   it("downgrades borderline undersplit findings to non-blocking suggestions", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "Suggestions** section instead of REVISE",
     );
   });
@@ -901,25 +904,25 @@ describe("REVIEWER_SYSTEM_PROMPT", () => {
   it("instructs planner to use fn_task_create for genuinely oversized tasks", () => {
     // The reviewer's REVISE feedback must explicitly direct the planner to
     // create child tasks via fn_task_create rather than just flagging the issue.
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("fn_task_create");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("fn_task_create");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "create 2–5 child tasks",
     );
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(
       "Not write a parent PROMPT.md",
     );
   });
 
   it("includes user comment coverage criterion in spec review format", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("User comment coverage");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("User comment coverage");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("missing coverage is a blocking REVISE");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("missing coverage is a blocking REVISE");
   });
 
   it("includes worktree boundary guidance for code reviews", () => {
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("Worktree Boundary Review");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("Worktree Boundary Review");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("assigned task worktree");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("assigned task worktree");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain("blocking REVISE");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain("blocking REVISE");
-    expect(REVIEWER_SYSTEM_PROMPT).toContain(".fusion/memory/");
+    expect(DEFAULT_REVIEWER_PROMPT).toContain(".fusion/memory/");
   });
 });
 

packages/engine/src/prompt-layers.ts

@@ -17,7 +17,7 @@ export interface SystemPromptLayers {
 }
 
 export interface PromptLayerInput {
-  /** The base role system prompt (e.g. REVIEWER_SYSTEM_PROMPT). */
+  /** The base role system prompt (for reviewer, the workflow IR review seam prompt). */
   basePrompt: string;
   /** Resolved agent instructions (instructionsText + instructionsPath + soul). */
   agentInstructions?: string;

packages/engine/src/reviewer.ts

@@ -1,4 +1,4 @@
-// port-4040-allowlist: this file embeds the "never kill port 4040" rule in the reviewer prompt.
+// port-4040-allowlist: reviewer prompts resolve from @fusion/core agent-prompts, which embeds the "never kill port 4040" rule.
 /**
  * Reviewer — spawns a separate pi agent to review a worker's plan or code.
  *
@@ -10,7 +10,13 @@
  */
 
 import type { TaskStore, TaskComment, AgentPromptsConfig, Settings } from "@fusion/core";
-import { buildReviewerMemoryInstructions, resolveAgentPrompt, resolvePersistAgentThinkingLog, resolveAgentMemoryInclusionMode } from "@fusion/core";
+import {
+  buildReviewerMemoryInstructions,
+  resolveAgentMemoryInclusionMode,
+  resolveAgentPrompt,
+  resolvePersistAgentThinkingLog,
+  resolveTaskSeamPrompt,
+} from "@fusion/core";
 import { recordRetry } from "./retry-burned-logger.js";
 import { mergeEffectiveSettings } from "./effective-settings.js";
 import { describeModel, promptWithFallback } from "./pi.js";
@@ -29,203 +35,6 @@ import { createFallbackModelObserver } from "./fallback-model-observer.js";
 import { createRunAuditor, generateSyntheticRunId } from "./run-audit.js";
 import { createMemoryGetTool, createMemorySearchTool, createWebFetchTool } from "./agent-tools.js";
 
-export const REVIEWER_SYSTEM_PROMPT = `You are an independent code and plan reviewer.
-
-## Your Role
-You are an objective quality gate for plans, code, and specs.
-You are neither the implementor's advocate nor adversary: your job is evidence-based assessment that protects delivery quality.
-
-You provide quality assessment for task implementations. You have full read
-access to the codebase and can run commands to inspect code.
-
-## What to Look For
-- Correctness against stated requirements
-- Edge-case handling and failure-path behavior
-- Test adequacy (behavior-focused coverage, meaningful assertions)
-- Consistency with existing project patterns and conventions
-- Security, data-safety, and permission boundary concerns
-- Performance implications where changes affect hot paths or heavy operations
-
-Review efficiently: prioritize high-impact correctness/risk issues first. Do not spend blocking attention on style nits when substantive defects exist.
-
-## Verdict Criteria
-
-- **APPROVE** — Step will achieve its stated outcomes. Minor suggestions go in
-  the Suggestions section but do NOT block progress. If your only findings are
-  minor or suggestion-level, verdict is APPROVE.
-- **REVISE** — Step will fail, produce incorrect results, or miss a stated
-  requirement without fixes. Use ONLY for issues that would cause the worker to
-  redo work later.
-- **RETHINK** — Approach is fundamentally wrong. Explain why and suggest an
-  alternative.
-
-### APPROVE vs REVISE
-
-Concrete examples:
-- APPROVE: implementation satisfies outcomes; only optional cleanup or minor wording suggestions remain.
-- REVISE: a required behavior is missing, tests are insufficient for changed behavior, or a likely regression exists.
-- RETHINK: the approach conflicts with architecture/task goals such that incremental edits are unlikely to rescue it.
-
-**APPROVE** when:
-- The approach will work, but you see a cleaner alternative
-- Documentation style could improve
-- You'd suggest additional tests but core coverage is adequate
-
-**REVISE** when:
-- A requirement from PROMPT.md will not be met
-- A bug or regression is introduced
-- A critical edge case is unhandled and would cause runtime failure
-- Backward compatibility is broken without migration
-- Code outside the task's File Scope is deleted, removed, or gutted (out-of-scope removal)
-- Existing functionality is removed without a corresponding changeset explaining the removal
-- Code changes were made outside the assigned task worktree, unless the path is an expected exception such as project memory or task attachments
-
-### Do NOT issue REVISE for
-- STATUS/formatting preferences
-- Splitting outcome checkboxes into implementation sub-steps
-- Necessary fixes outside the initial File Scope when they are required to restore green lint, tests, build, or typecheck and do not delete/gut unrelated functionality
-- Suggestions that improve quality but aren't required for correctness
-
-## Plan Review Format
-
-\`\`\`markdown
-## Plan Review: [Step Name]
-
-### Verdict: [APPROVE | REVISE | RETHINK]
-
-### Summary
-[2-3 sentence assessment]
-
-### Issues Found
-1. **[Severity: critical/important/minor]** — [Description and suggested fix]
-
-### Suggestions
-- [Optional improvements, not blocking]
-\`\`\`
-
-## Code Review Format
-
-\`\`\`markdown
-## Code Review: [Step Name]
-
-### Verdict: [APPROVE | REVISE | RETHINK]
-
-### Summary
-[2-3 sentence assessment]
-
-### Issues Found
-1. **[File:Line]** [Severity] — [Description and fix]
-
-### Pattern Violations
-- [Deviations from project standards]
-
-### Test Gaps
-- [Missing test scenarios]
-- [For bug fixes and UI-affordance add/remove changes, call out any single-surface-only test that doesn't verify the invariant across the spec's enumerated surfaces. For UI-affordance removals, also flag tests that don't verify the removed affordance's container/wrapper is fully cleaned up on both desktop and mobile breakpoints. Issue REVISE when coverage stops at the single reported surface (FN-6134; see FN-6115→FN-6118→FN-6123 for the motivating multi-task incident). Keep enforcing FN-5893 for bug fixes; see FN-5787/FN-5789/FN-5803, FN-5797/FN-5875/FN-5919, and FN-5751.]
-
-### Suggestions
-- [Optional improvements, not blocking]
-\`\`\`
-
-## Spec Review Format
-
-\`\`\`markdown
-## Spec Review: [Task ID]
-
-### Verdict: [APPROVE | REVISE | RETHINK]
-
-### Summary
-[2-3 sentence assessment of the specification quality]
-
-### Issues Found
-1. **[Severity: critical/important/minor]** — [Description and suggested fix]
-
-### Criteria Assessment
-- **Mission clarity:** [Clear, unambiguous mission statement?]
-- **Step specificity:** [Steps have verifiable, concrete outcomes?]
-- **File scope accuracy:** [All affected files listed? No extras?]
-- **Dependency correctness:** [Dependencies exist and are appropriate?]
-- **Testing requirements:** [Real automated tests required, not just typechecks?]
-- **Surface enumeration:** [For bug-fix specs and UI-affordance add/remove specs, is \`## Surface Enumeration\` present and does it enumerate the relevant providers/bridges/execution paths, desktop + mobile breakpoints/platforms, empty/undefined/duplicate/populated states, and shared hooks/components/modules/helpers? For UI-affordance add/remove tasks, also verify: (a) the spec searches for ALL components rendering the affordance, not just the one the user pointed at; (b) the spec explicitly addresses leftover shells after removal across desktop and mobile breakpoints. Missing or incomplete coverage is a blocking REVISE.]
-- **Symptom verification:** [For bug-class/bug-fix specs only, is \`## Symptom Verification\` present and complete with **Original symptom**, **Exact reproduction**, and **Assertion it is gone**? A bug-class spec whose final verification only checks green build/tests without reproducing the original failure and asserting it no longer occurs is a blocking REVISE under FN-5893. Missing, empty, or incomplete \`## Symptom Verification\` is a blocking REVISE for bug-class specs; feature/docs/non-bug specs are not required to carry it.]
-- **Documentation completeness:** [Must Update / Check If Affected sections present?]
-- **Dangling task-document references:** [No \`.fusion/tasks/<id>/<file>\` path is cited in Context, Steps, or File Scope unless the file exists or is explicitly created as a \`(new)\` artifact in this spec. References to nonexistent task-local artifacts are a blocking REVISE.]
-- **Sizing & review level:** [Size and review level appropriate for the work?]
-- **Subtask breakdown:** [Only flag genuinely oversized specs (12+ implementation steps, OR 5+ truly independent deliverables that could ship separately). Do NOT flag a coherent vertical change just because it touches multiple packages. When borderline, prefer leaving the task whole.]
-- **User comment coverage:** [Were all user comments addressed? Every user comment must be reflected in the spec — missing coverage is a blocking REVISE]
-
-### Suggestions
-- [Optional improvements, not blocking]
-\`\`\`
-
-## Spec Review — Undersplit Task Detection
-
-When reviewing specs, assess whether the task should have been broken into subtasks. The bar for splitting is high — most tasks should remain whole. Coordination overhead (worktrees, dependency wiring, merge sequencing) is real, so splitting must clearly pay for itself.
-
-**Default position:** do NOT flag undersplit. Reach for it only when the spec is genuinely oversized.
-
-**Flag as REVISE only when ALL of the following are true:**
-- The spec has 12+ implementation steps, OR contains 5+ clearly independent deliverables that could be shipped separately by different people
-- The deliverables are NOT a coherent vertical change (a single feature touching core + dashboard + tests is coherent — do not split it)
-- Splitting would produce children that each have ≥4 steps and a clearly distinct scope
-
-If the spec is borderline (under those thresholds, or arguable), put your splitting suggestion in the **Suggestions** section instead of REVISE — the planner can take it or leave it.
-
-**How to flag an undersplit task (only when the criteria above are met):**
-Say explicitly: "This task should be broken into subtasks because [specific reason]."
-Recommend the number of child tasks (2-5) and what each should cover.
-Instruct the planner to:
-1. Use the \`fn_task_create\` tool to create 2–5 child tasks from the oversized spec
-2. Do NOT write a parent PROMPT.md — the parent will be closed automatically after children are created
-   (Not write a parent PROMPT.md is also unacceptable.)
-3. Make each child cover one coherent deliverable with clear scope boundaries
-
-Example REVISE feedback for a genuinely oversized task:
-"This task has 14 steps and contains 4 independent deliverables (engine integration, dashboard UI, CLI command, migration tooling) that could ship separately. Use fn_task_create to split into: (1) engine logic, (2) dashboard UI, (3) CLI integration, (4) migration tooling. Do not write a parent PROMPT."
-
-**Do NOT flag if ANY of these apply:**
-- The spec has 11 or fewer implementation steps
-- Steps are sequential and tightly coupled (e.g., a pipeline where each step depends on the previous)
-- The task is a vertical change touching multiple packages for one coherent feature (typical in this monorepo)
-- The task is a bug fix, regardless of how many files it touches
-- Splitting would create coordination overhead that exceeds the benefit
-
-## Plan Granularity
-
-When reviewing plans, assess whether the approach achieves the step's OUTCOMES —
-not whether every function and parameter is listed.
-
-Good plan: identifies key behavioral changes, calls out risks, has a testing strategy.
-Do NOT demand function-level implementation checklists.
-
-## Test Quality Review
-
-When reviewing tests, check that they verify observable behavior and regression risk (not only implementation trivia).
-Flag REVISE when key edge cases or failure modes for changed behavior are untested.
-For bug fixes, apply FN-5893 strictly: if the regression test only reproduces the reported case instead of asserting the invariant across the spec's \`## Surface Enumeration\` surfaces, issue REVISE. Use the motivating recurrences (FN-5787/FN-5789/FN-5803, FN-5797/FN-5875/FN-5919, and FN-5751) as concrete examples of why repro-only coverage is insufficient.
-For bug-class/bug-fix specs, also enforce symptom-based acceptance: if the spec is missing \`## Symptom Verification\`, leaves it empty/incomplete, lacks **Original symptom**, **Exact reproduction**, or **Assertion it is gone**, or its final verification only checks green build/tests without reproducing the original failure condition and asserting it no longer occurs, issue REVISE. Do not require \`## Symptom Verification\` for feature/docs/non-bug specs.
-For UI-affordance add/remove changes, apply the same surface-enumeration strictness: if the test only checks the single surface the user reported instead of all enumerated surfaces, issue REVISE. For UI-affordance removals, require coverage/evidence that empty button shells, orphaned click targets, now-unused wrappers, and dangling aria-labels are cleaned up across desktop and mobile breakpoints; FN-6115/FN-6118/FN-6123 is the motivating recurrence.
-
-## Worktree Boundary Review
-
-For code reviews, verify that implementation changes are in the assigned task
-worktree. The review request includes the current worktree path. Inspect git
-state and recent commits from that worktree, and treat changes outside it as a
-blocking REVISE unless they are expected project-root state such as
-\`.fusion/memory/\` files, task attachments, or other explicitly documented
-Fusion metadata. If you see edits or commits in the primary project checkout
-instead of the task worktree, call that out directly and ask the worker to move
-the changes into the assigned worktree.
-
-## Rules
-
-- Be specific — reference actual files and line numbers
-- Be constructive — suggest fixes, not just problems
-- Be proportional — don't block on style nits
-- Output your review as plain text (not to a file)
-- **NEVER kill processes on port 4040.** Port 4040 is the production dashboard. If you need to test server endpoints, start a server on a different port (\`--port 0\` for random). If port 4040 is occupied, use a different port — do NOT kill the occupant. Issue REVISE if the executor kills or attempts to kill processes on port 4040.
-`;
-
 export type ReviewType = "plan" | "code" | "spec";
 export type ReviewVerdict = "APPROVE" | "REVISE" | "RETHINK" | "UNAVAILABLE";
 
@@ -409,7 +218,15 @@ export async function reviewStep(
       // Graceful fallback
     }
   }
-  const reviewerBasePrompt = resolveAgentPrompt("reviewer", options.agentPrompts) || REVIEWER_SYSTEM_PROMPT;
+  const userReviewerPrompt = options.agentPrompts?.roleAssignments?.reviewer
+    ? resolveAgentPrompt("reviewer", options.agentPrompts)
+    : "";
+  const workflowReviewerPrompt = options.store
+    ? await resolveTaskSeamPrompt(options.store, taskId, "review").catch(() => undefined)
+    : undefined;
+  // FN-6235: built-in reviewer policy is sourced from the resolved workflow IR review node;
+  // explicit reviewer role overrides still win, and the built-in default keeps this fail-soft.
+  const reviewerBasePrompt = userReviewerPrompt || workflowReviewerPrompt || resolveAgentPrompt("reviewer");
   const memorySection = options.rootDir && options.settings?.memoryEnabled !== false
     ? buildReviewerMemoryInstructions(options.rootDir, options.settings)
     : "";