Review Implementation

Overview

Review code against the given markdown docs, identify gaps and risks, refactor as needed, and apply fixes while keeping changes minimal and aligned to the requirements.

Workflow

1) Intake and scope

• Read all provided .md files and extract explicit requirements, constraints, and acceptance criteria.
• Note implied expectations and any ambiguity; ask a question only if a critical decision cannot be made safely.
• Check for repo-specific guidelines (limits, style, tests) and follow them.

2) Map requirements to implementation

• Locate the relevant code paths and build a checklist mapping each requirement to code evidence.
• Flag missing, partial, or conflicting implementation.

3) Review design, quality, risk, and scope

• Identify critical issues first: correctness, security, data loss, concurrency, API contract breaks, and major performance regressions.
• Evaluate design soundness: module boundaries, responsibility split, data flow, coupling, and long-term maintainability.
• Detect over-implementation: extra features, unused abstractions, or complexity beyond the docs.
• Find code health issues: overly long files/functions, duplicated logic, unclear naming, or unnecessary indirection.

4) Apply fixes and refactors

• Fix issues directly in code with minimal, targeted changes.
• Split overly long files/functions into smaller units with clear, intention-revealing names.
• Remove or simplify overbuilt code that is not justified by the docs.
• Keep behavior consistent with the documented requirements.

5) Validate and report

• Run relevant tests/linters if available and allowed; otherwise state what was not run.
• Report findings in severity order with file/line references, then list fixes applied and any remaining risks or decisions made.
• For non-critical decisions, choose the optimal approach and proceed; for critical blockers, clearly flag them and ask.