---
name: implementation-plan-review
description: Review implementation tasks for coverage, feasibility, and multi-model validation
summary: "Verifies every feature has implementation tasks, no task is too large for one session, the dependency graph has no cycles, and every acceptance criterion maps to at least one task."
phase: "planning"
order: 1220
dependencies: [implementation-plan]
outputs: [docs/reviews/review-tasks.md, docs/reviews/implementation-plan/task-coverage.json, docs/reviews/implementation-plan/review-summary.md, docs/reviews/implementation-plan/codex-review.json, docs/reviews/implementation-plan/antigravity-review.json]
conditional: null
knowledge-base: [review-methodology, review-implementation-tasks, task-decomposition, multi-model-review-dispatch, review-step-template]
---

## Purpose
Review implementation tasks targeting task-specific failure modes: architecture
coverage gaps, missing dependencies, tasks too large or too vague for agents,
agent executability violations, critical path inaccuracy, and invalid
parallelization assumptions. At depth 4+,
dispatch to independent AI models (Codex/Antigravity CLIs) for multi-model validation
and produce a structured coverage matrix and review summary.

## Inputs
- docs/implementation-plan.md (required) — tasks to review
- docs/system-architecture.md (required at deep; optional — not available in MVP) — for coverage checking
- docs/domain-models/ (required at deep; optional — not available in MVP) — for completeness
- docs/user-stories.md (required) — for AC coverage mapping
- docs/plan.md (required) — for traceability
- docs/project-structure.md (required) — for file contention analysis
- docs/tdd-standards.md (required) — for test requirement verification

## Expected Outputs
- docs/reviews/review-tasks.md — findings and resolution log
- docs/implementation-plan.md — updated with fixes
- docs/reviews/implementation-plan/task-coverage.json — AC-to-task coverage matrix (depth 3+)
- docs/reviews/implementation-plan/review-summary.md — multi-model review summary (depth 4+)
- docs/reviews/implementation-plan/codex-review.json — raw Codex findings (depth 4+, if available)
- docs/reviews/implementation-plan/antigravity-review.json — raw Antigravity findings (depth 4+, if available)

## Quality Criteria
- (deep) Architecture coverage verified (every component has tasks)
- (mvp) Every user story maps to at least one implementation task
- (mvp) Dependency graph is valid DAG
- (mvp) No task is too large for a single agent session
- (deep) Critical path is accurate
- (deep) Parallelization assumptions are valid
- (deep) Every acceptance criterion maps to at least one task (100% AC coverage)
- (deep) Every task has verb-first description, >= 1 input file reference, >= 1 acceptance criterion, and defined output artifact
- (mvp) Every task complies with agent executability rules (3-file, 150-line, single-concern, decision-free, test co-location)
- (mvp) Tasks exceeding limits have explicit `<!-- agent-size-exception -->` justification
- (mvp) Plan Output Contract holds: every task and container has an ID; IDs are unique and stable; all parent and `depends_on` refs resolve; the dependency graph is acyclic; field blocks parse
- (depth 4+) Independent model reviews completed and reconciled

## Plan Output Contract Validation

The plan is consumed by an automated materializer that needs stable join keys
and a parseable structure (see the **Plan Output Contract** section in
`implementation-plan.md`). This review MUST verify that contract holds, rejecting
any plan that violates it:

- **ID presence** — every task has a `T-NNN` ID, and (deep) every story has an
  `S-NNN` ID and every epic an `E-NNN` ID. Flag any item missing an ID.
- **Uniqueness and stability** — task, story, and epic IDs are each **unique**
  and **stable** (no two items share an ID; no ID changed for an item that
  already existed in a prior plan revision; IDs are never reused for a different
  task/container).
- **Referential integrity (no dangling refs)** — every `story` / `epic` parent
  reference **and every `depends_on` reference** **resolves** to a declared ID in
  the plan. Report any **dangling** ref (a parent or `depends_on` entry that does
  not resolve to a declared ID) as a blocking finding. A story with no `epic`
  parent is valid, not dangling.
- **Acyclicity** — the dependency graph formed by all `depends_on` edges is a
  **DAG**: there are **no cycles**. Walk the graph and report any cycle.
- **Parseability** — every per-task and per-container fenced metadata block
  parses cleanly under the canonical serialization (well-formed key/value block
  with the required fields present).

## Methodology Scaling
- **deep**: Full multi-pass review with multi-model validation. AC coverage
  matrix. Independent Codex/Antigravity dispatches. Detailed reconciliation report.
- **mvp**: Coverage check only. No external model dispatch.
- **custom:depth(1-5)**:
  - Depth 1: architecture coverage check (every component has tasks).
  - Depth 2: coverage check plus DAG validation and agent executability rules.
  - Depth 3: add dependency analysis, AC coverage matrix, and task sizing audit.
  - Depth 4: add one external model review (Codex or Antigravity).
  - Depth 5: full multi-model review with reconciliation and detailed findings report.

## Mode Detection
Re-review mode if previous review exists. If multi-model review artifacts exist
under docs/reviews/implementation-plan/, preserve prior findings still valid.

## Update Mode Specifics

- **Detect**: `docs/reviews/review-tasks.md` exists with tracking comment
- **Preserve**: Prior findings still valid, resolution decisions, multi-model review artifacts
- **Triggers**: Upstream artifact changed since last review (compare tracking comment dates)
- **Conflict resolution**: Previously resolved findings reappearing = regression; flag and re-evaluate