# Pc Implementation Alignment Review

> Use after implementation or repair work when the reviewer must judge intent realization, requirement coverage, scope consistency, and claim truth before code-review or delivery-completion can accept the work.

- Skill: `yknothing/pc-implementation-alignment-review` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add yknothing/pc-implementation-alignment-review`
- Raw SKILL.md: https://api.skillmd.com/api/skills/yknothing/pc-implementation-alignment-review/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: yknothing (https://skillmd.com/u/yknothing)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/yknothing/pc-implementation-alignment-review

---


# Implementation Alignment Review

## Context

This skill checks whether the implementation actually satisfies the user's intent and approved scope.

See [context notes](references/context.md).

## Inputs

[I/O contract notes](references/io-contract.md) define required inputs and authority.

## Process

### Step 1: Reconstruct the Intent Contract

State the intended outcome in one or two concrete sentences. Separate must-have behavior, non-targets, and later work. Do not infer extra scope from implementation shape.

### Step 2: Build a Coverage Matrix

Map each user requirement, acceptance criterion, and priority item to implementation evidence. Mark each row as implemented, partially implemented, deferred, contradicted, or unverifiable.

### Step 3: Compare Behavior to Claims

Check final messages, docs, quality reports, release notes, and test names against the actual code and runtime evidence. Flag any claim that turns simulated, fake, fixture, local-only, or partial evidence into a stronger statement.

### Step 4: Check Consistency Across Artifacts

Verify that code, tests, docs, runbooks, risk registers, and quality snapshots tell the same story. Mismatches are findings even when each file is locally coherent.

### Step 5: Classify Gaps

Prioritize:

- **Blocking**: Intent not implemented, wrong target, false completion claim, missing acceptance criterion, or unowned contradiction.
- **Should-fix**: Partial coverage or ambiguous evidence that could mislead a downstream reviewer.
- **Question**: Missing context that can change severity.

## Outputs

Produce only declared outputs at their documented quality boundary.

## Quality Gate

- [ ] Every must-have intent item maps to code, test, documentation, or an explicit deferred/non-target decision.
- [ ] No simulated, fake, fixture, or local-only evidence is described as real runtime or production evidence.
- [ ] Completion status is no stronger than the weakest required evidence.
- [ ] Cross-file contradictions are either fixed or reported as findings.