Code review Extension for Spec Kit

Post-implementation code review with specialized agents for code quality, comments, tests, error handling, type design, and simplification. Orchestrates 6 focused review agents into a single consolidated report with severity-based grouping and actionable remediation guidance.

Features

• Coordinator (/speckit.review.run): Orchestrates all agents, produces a consolidated report
• Code Reviewer (/speckit.review.code): Project guideline compliance, bug detection, code quality
• Comment Analyzer (/speckit.review.comments): Comment accuracy, documentation completeness, comment rot
• Test Analyzer (/speckit.review.tests): Behavioral coverage, critical gap identification, test resilience
• Error Handling Reviewer (/speckit.review.errors): Silent failure detection, catch block analysis, error logging
• Type Design Analyzer (/speckit.review.types): Encapsulation, invariant expression, usefulness, and enforcement
• Code Simplifier (/speckit.review.simplify): Clarity analysis, unnecessary complexity, redundant abstractions

Installation

# From community catalog
specify extension add review

# Or from repository directly
specify extension add review --from https://github.com/ismaelJimenez/spec-kit-review/archive/refs/tags/v1.0.1.zip

# Local development
specify extension add --dev /path/to/spec-kit-review

Verify installation:

specify extension list
# Should show:
#  ✓ Review Extension (v1.0.1)
#     Post-implementation comprehensive code review with specialized agents for code quality, comments, tests, error handling, type design, and simplification.
#     Commands: 7 | Hooks: 1 | Status: Enabled

Usage

Full Coordinated Review

Run all specialized agents against your changes and get a consolidated report:

/speckit.review.run

All commands (coordinator and individual agents) use the built-in detect-changed-files script to automatically identify what to review when no files are specified:

• Feature branch: Committed changes since the merge base with the default branch (main/master), plus any uncommitted work
• Default branch: Only uncommitted work (staged and unstaged changes)

You can skip the script entirely by telling the agent what to review:

/speckit.review.run only staged changes
/speckit.review.run only files in src/utils/

Targeted Review

Run only specific agents by passing aspect names:

/speckit.review.run tests errors      # Only test and error handling analysis
/speckit.review.run code              # Only code quality review
/speckit.review.run comments simplify # Only comment analysis and simplification

Valid aspects: code, comments, tests, errors, types, simplify, all

Parallel Review

By default agents run sequentially so you can act on each report as it arrives. Add parallel to launch all agents simultaneously for faster results:

/speckit.review.run all parallel      # Full review, all agents in parallel
/speckit.review.run tests errors parallel  # Parallel targeted review

Parallel mode is useful for comprehensive reviews where you want all findings at once rather than incremental feedback.

Direct Agent Invocation

Run any agent directly for focused, deep analysis:

/speckit.review.code       # Code quality review
/speckit.review.comments   # Comment accuracy analysis
/speckit.review.tests      # Test coverage analysis
/speckit.review.errors     # Error handling review
/speckit.review.types      # Type design analysis
/speckit.review.simplify   # Code simplification suggestions

Each agent auto-detects changed files independently when invoked directly.

Report Output

The consolidated report includes:

• Critical Issues: Must-fix issues identified by agents — file, line, description
• Important Issues: Should-fix issues — file, line, description
• Suggestions: Nice-to-have improvements — file, line, description
• Strengths: What's well-done in the PR
• Recommended Action: Prioritized remediation steps

Configuration

Project Guidelines

If project-specific guidelines exist (.specify/memory/constitution.md, CLAUDE.md, .github/copilot-instructions.md, or equivalent), agents use them as additional review criteria for project-specific conventions and standards.

Environment Requirements

• git: Required for change detection
• spec-kit: >= 0.1.0

Token Usage

Heads up: A full coordinated review (/speckit.review.run) dispatches 6 specialized agents, each of which reads the changed files independently. This can be token-intensive on larger PRs. To reduce costs, run targeted reviews (/speckit.review.run code errors) instead of the full suite.

Recommended Workflow

1. Implement changes:       /speckit.implement
2. Run full review:         /speckit.review.run
3. Fix critical issues
4. Re-run targeted review:  /speckit.review.run code errors
5. Verify fixes resolved
6. Create PR

Integration with Verify Extension

If you also use the Verify Extension (spec-kit-verify), the recommended workflow is:

1. Implement changes:       /speckit.implement
2. Verify spec alignment:   /speckit.verify.run
3. Run PR review:           /speckit.review.run
4. Fix issues and iterate

The verify extension validates that your implementation matches specification artifacts (spec.md, plan.md, tasks.md). The review extension then performs broader code quality analysis. When both are installed, the verify extension offers a handoff to run the review automatically after verification completes.

Troubleshooting

Issue: Command not available

Solutions:

1. Check extension is installed: specify extension list
2. Restart AI agent
3. Reinstall extension: specify extension add review

Issue: Validation Error: Invalid alias 'speckit.review'

Solution: Upgrade to v1.0.1 or later and invoke the coordinator with /speckit.review.run. Spec Kit now enforces three-segment alias names, so /speckit.review is no longer accepted by the validator.

Issue: "Not a git repository" error

Solution: The review extension requires git for change detection. Initialize a git repository with git init or run commands from within an existing repo.

Issue: "No changes detected"

Solution: Make some code changes first. On a feature branch, commit changes. On the default branch, stage or modify files.

Acknowledgments

The first version of this extension was modeled after the PR Review Toolkit plugin for Claude Code by Anthropic.

License

MIT License — see LICENSE file

Support

• Issues: https://github.com/ismaelJimenez/spec-kit-review/issues
• Spec Kit Docs: https://github.com/github/spec-kit

Changelog

See CHANGELOG.md for version history.

Extension Version: 1.0.1 · Spec Kit: >=0.1.0