Review pull requests and diffs like a senior engineer.

Use when the user asks to review a PR, check code quality, analyze changes, or provide feedback on a branch or diff. Reads the actual code changes, flags real issues, groups feedback by severity, and never rubber-stamps.

Write clean, reviewable git commits.

Use when the user asks to commit, stage changes, split work into commits, write or fix a commit message, or prepare a branch for a pull request. Follows the repository's own conventions first, then Conventional Commits.

Write a clear pull request title and description from the branch's actual changes.

Use when the user asks to open, draft, or describe a pull request, fill in a PR template, or summarise a branch for review. Follows the repository's PR template and title rules, and never opens a PR without being asked.

Write a changelog entry for a release from commits, tags, and diffs.

Use when the user asks to update CHANGELOG.md, write release notes, cut a version, or summarise what changed since the last tag. Follows Keep a Changelog format by default, respects the repo's own format when it exists, and never bumps the version or publishes without being asked.

Diagnose and fix bugs systematically.

Use when the user reports an error, unexpected behaviour, a failing test, or a crash. Reads logs and code, reproduces the issue, isolates the root cause, fixes it, and verifies the fix — without guessing or changing code blindly.

Refactor code safely with no behaviour changes.

Use when the user asks to clean up, simplify, rename, extract, modularise, split a large file, add logging, improve comments, or reduce complexity. Verifies tests pass before and after every change, and never mixes refactoring with features or bug fixes.

Design clean, consistent HTTP APIs.

Use when the user asks to design an endpoint, review an API contract, add a new route, define request/response shapes, write an OpenAPI spec, or decide on REST vs GraphQL vs RPC for a feature.

Audit outdated dependencies, review changelogs for breaking changes, update packages safely, and write a clean commit.

Use when the user asks to update dependencies, upgrade packages, check for outdated deps, or bump a library version. Never upgrades a major version blindly — always reads the changelog first and surfaces breaking changes before touching any file.

Get a project running locally from a clean clone — installs the right runtime, dependencies, and config, then confirms the baseline build and tests pass.

Use when the user asks to set up a project, get it running, onboard to a repo, or fix a broken local environment. Reads the project's own docs and config files first; never assumes a stack without checking.

Orchestrate a full release: bump the version, update the changelog, create a git tag, and draft release notes — ready to push when the user says so.

Use when the user asks to cut a release, tag a version, ship a release, or publish a new version. Never pushes, publishes, or deploys without being explicitly asked.

Write safe, readable, and performant SQL.

Use when the user asks to write a query, optimise a slow query, add an index, design a schema, review SQL for correctness or injection risk, or translate application logic into SQL.

Write meaningful tests for existing or new code.

Use when the user asks to add tests, increase coverage, write unit or integration tests, test an edge case, or set up a test suite. Focuses on tests that catch real bugs, not tests that inflate a coverage number.

Diagnose and fix service startup race conditions in containerized environments.

Use when a CI job fails with a service that reports healthy but behaves broken, a container passes its healthcheck but rejects writes or connections, a dependent service says "ready" but the dependent component crashes, or a distroless image silently breaks a healthcheck command.

Write, review, and fix Dockerfiles, Compose files, and container build configurations.

Use when the user asks to write a Dockerfile, containerize an app, fix a slow build, reduce image size, write a docker-compose.yml, set up a multi-service stack, add a healthcheck, fix a build error, or debug a container that won't start.

Reduce unnecessary token consumption in AI coding-agent sessions without losing technical accuracy, implementation quality, or essential evidence.

Use when the user says 'save tokens', 'be more concise', 'use fewer tokens', 'stop repeating yourself', 'tokenjars', or when a session is growing long and the agent is repeating itself, re-reading unchanged files, or producing verbose tool-call narration. Works in Balanced (default), Strict, or Minimal mode.

Write, review, and fix CI/CD pipeline configurations for GitHub Actions, GitLab CI, and similar platforms.

Use when the user asks to add a CI pipeline, set up GitHub Actions, write a workflow file, fix a failing pipeline, add a deploy step, set up staging or production deploys, wire up secrets, add branch protection, or speed up a slow pipeline.

Audit and clean up code quality issues that accumulate over time — not bugs, but rot.

Use when the user asks to clean up code, fix naming conventions, remove unused imports, remove dead code, fix inconsistent variable names, eliminate AI-generated boilerplate, remove vendor-generated noise, standardize casing (camelCase, snake_case, PascalCase), remove stale comments, fix magic numbers, or sweep a file or module for code quality issues before a release or review.

Profile, diagnose, and fix performance bottlenecks using the right tools for the stack.

Use when the user says something is slow, asks why an endpoint is taking too long, wants to reduce memory usage, needs to find what is hogging CPU, says a query is slow, asks to profile the app, or wants to improve response time. Never guesses — always measures before changing code.

Audit and fix web accessibility issues to meet WCAG 2.1 AA.

Use when the user asks to make a component accessible, fix accessibility errors, add ARIA attributes, improve keyboard navigation, check color contrast, make a form accessible, fix screen reader issues, or audit a page for a11y problems.

Add, improve, and fix structured logging in an application.

Use when the user asks to add logging, fix noisy logs, improve log output, add structured logs, set up log levels, replace console.log with a real logger, add request tracing, or make logs parseable by a log aggregator like Datadog, CloudWatch, or Loki.

Write, review, and run database migrations safely.

Use when the user asks to add a migration, rename a column, drop a table, change a schema, run migrations, roll back a migration, write a zero-downtime schema change, or fix a failed migration. Never runs destructive migrations without confirmation. Always reads the current schema before writing a migration.

Generate, validate, and maintain OpenAPI specs for HTTP APIs.

Use when the user asks to write an OpenAPI spec, generate a swagger file, document an API, add request/response schemas, validate an existing spec, keep the spec in sync with code, add examples to an OpenAPI file, or fix OpenAPI validation errors.

Implement and review authentication and authorization in web applications.

Use when the user asks to add login, implement JWT, set up sessions, add OAuth, protect a route, implement role-based access control (RBAC), fix an auth bug, or review auth code for security issues. Never cuts corners on security — always reads the existing auth setup before adding to it.

Add caching to reduce latency and database load.

Use when the user asks to add a cache, speed up repeated queries, set up Redis, implement cache invalidation, add HTTP caching headers, use a CDN, fix stale cache bugs, or decide what to cache. Never caches blindly — measures first and picks the right strategy for the data.

Add and improve error handling in web applications and APIs.

Use when the user asks to improve error handling, add a global error handler, create custom error classes, fix unhandled promise rejections, return consistent error responses, stop leaking stack traces to clients, handle async errors, or make errors easier to debug in production.

Add rate limiting to protect APIs and services from abuse, brute-force attacks, and traffic spikes.

Use when the user asks to add rate limiting, prevent brute-force login attempts, throttle an API endpoint, implement a token bucket or sliding window, add per-user or per-IP limits, or fix a service that is getting hammered by excessive requests.

Implement WebSocket connections for real-time features.

Use when the user asks to add real-time updates, implement a chat feature, push live notifications, add a live dashboard, set up Socket.IO, use the WebSocket API, implement pub/sub over a socket, or scale WebSockets across multiple server instances.

>

>

>

>

>

>

>

>

>

>

>

>