Skip to content
All Skills

Review Merge Worker

Use when an evaluation run must merge research, findings, documentation, and repair outputs into one verified result.

Software Engineering|v1|Updated 5/18/2026|GitHub source
MCP get_skill({ skillId: "review-merge-worker-1cd4654e" })

Use this skill with your agent

Create a free account and connect via MCP

Get Started Free
> **Paths:** File paths (`references/`, `../ln-*`) are relative to this skill directory.

**Type:** L3 Worker
**Category:** 3XX Planning

# Review Merge Worker

## Mandatory Read

**MANDATORY READ:** Load `references/evaluation_worker_runtime_contract.md`, `references/evaluation_summary_contract.md`

## Purpose

- merge read-only evidence lanes after their join barrier
- deduplicate overlap against prior review history
- produce one verified aggregate result for the coordinator

## Wait/Patience Protocol

This worker does not launch agents. It consumes worker summaries and existing agent result artifacts after the evaluation runtime barrier. Do not mark an agent failed from elapsed time alone; accept only the runtime's resolved status or result metadata as the completion signal.

## Agent Finding Verification

For every agent suggestion:
- treat transport, auth, permission, timeout-without-output, or tool-missing results as operator evidence, not domain findings
- verify supported claims against code, docs, tests, runtime logs, or worker summaries before accepting
- reject unsupported, stale, duplicate, or architecture-shim suggestions
- never rewrite runner-owned result files

## Runtime

Runtime family:
- `evaluation-worker-runtime`

Required manifest fields:
- `identifier`
- `phase_order`
- `summary_kind=review-merge`
- `operation=merge`

Recommended `phase_order`:
1. `PHASE_0_CONFIG`
2. `PHASE_1_LOAD_WORKER_RESULTS`
3. `PHASE_2_DEDUPLICATE_AND_VERIFY`
4. `PHASE_3_WRITE_SUMMARY`
5. `PHASE_4_SELF_CHECK`

## Workflow

### Phase 1: Load Worker Results

1. Load all input worker summaries (ln-311, ln-312, ln-313, ln-314).
2. Load agent findings from all launched external agents.
3. Load `.hex-skills/agent-review/review_history.md` for prior review entries.

### Phase 2: Deduplicate and Verify

1. Deduplicate current findings against:
   - own analysis
   - worker findings
   - agent findings
   - prior review history
2. For each agent suggestion: independently verify per the Agent Finding Verification policy.
3. Mark each suggestion `AGREE` or `REJECT`.
4. **Architecture Gate:** Before accepting any AGREE'd suggestion, verify: "Does this implement the correct architecture directly, without backward compatibility shims or legacy workarounds?" If a suggestion introduces unnecessary compat layers, convert AGREE to REJECT.
5. Reject unsupported findings.

### Phase 3: Write Summary

Emit `summary_kind=review-merge`.

Payload must include:
- `worker=ln-315`
- `status`
- `operation=merge`
- `warnings`

Prefer these fields when available:
- `merge_summary.accepted_count`
- `merge_summary.rejected_count`
- `merge_summary.dedup_removed`
- `merge_summary.architecture_gate_rejections`

Save updated review summary to `.hex-skills/agent-review/review_history.md`.

### Phase 4: Self-Check

1. Verify deduplication completed.
2. Verify architecture gate applied to all AGREE'd suggestions.
3. Record `pass=true` only after summary write.

## Definition of Done

- [ ] Input worker summaries loaded
- [ ] Agent results loaded and verified per Critical Verification
- [ ] Duplicates removed (against findings + review history)
- [ ] Architecture Gate applied to all accepted suggestions
- [ ] Review summary saved to `review_history.md`
- [ ] `review-merge` summary written
- [ ] Self-check passed

**Version:** 1.0.0
**Last Updated:** 2026-04-10
#github#external#license-mit#levnikolaevich-claude-code-skills#agile-workflow#code-review#code#review

Related Skills

More skills in Software Engineering

Accessibility Standards

Comprehensive web accessibility standards based on WCAG 2.2 AA, with 38+ anti-patterns, legal enforcement context (EAA, ADA Title II), WAI-ARIA patterns, and framework-specific fixes for modern web frameworks and libraries.

#github-copilot#accessibilityMIT

Accord

Authoring unified specification packages across Business/Development/Design teams via staged elaboration (L0 Vision → L1 Requirements → L2 Team Detail → L3 Acceptance Criteria). No code. Use when authoring cross-team specs, building L0-L3 packages, or aligning Biz/Dev/Design on a single source of truth.

#broad-capability#developmentMIT

Acquire Codebase Knowledge

Use this skill when the user explicitly asks to map, document, or onboard into an existing codebase. Trigger for prompts like "map this codebase", "document this architecture", "onboard me to this repo", or "create codebase docs". Do not trigger for routine feature implementation, bug fixes, or narrow code edits unless the user asks for repository-level discovery.

#github-copilot#documentationMIT

Acreadiness Assess

Run the AgentRC readiness assessment on the current repository and produce a static HTML dashboard at reports/index.html. Wraps `npx github:microsoft/agentrc readiness` and hands off rendering to the @ai-readiness-reporter custom agent. Supports policies (--policy) for org-specific scoring. Use when asked to assess, audit, or score the AI readiness of a repo.

#github-copilot#planningMIT

Acreadiness Generate Instructions

Generate tailored AI agent instruction files via AgentRC instructions command. Produces .github/copilot-instructions.md (default, recommended for Copilot in VS Code) plus optional per-area .instructions.md files with applyTo globs for monorepos. Use after running /acreadiness-assess to close gaps in the AI Tooling pillar.

#github-copilot#skillMIT

Acreadiness Policy

Help the user pick, write, or apply an AgentRC policy. Policies customise readiness scoring by disabling irrelevant checks, overriding impact/level, setting pass-rate thresholds, or chaining org baselines with team overrides. Use when the user asks about strict mode, AI-only scoring, custom weights, CI gating, or wants org-wide standardisation.

#github-copilot#planningMIT