Personal Agent Workflows — an AI SDLC starter kit with 18 agents, 7 skills, 4 hooks, 6 rules, and a CLI.
This repository also configures its own agents. See what paw tells them →
Agent Claude Code
Produces a system architecture for a spec — components, boundaries, data flow, risks, alternatives. Use when a non-trivial change needs design before planning.
Agent Claude Code
Hunts latent bugs — race conditions, off-by-one, null deref, unbounded loops, integer overflow, encoding mismatches, time-of-check/time-of-use.
Agent Claude Code
Implements a task using TDD — writes the test first, then the code, then refactors. The only agent with full access.
Agent Claude Code
General code review on changed files. Flags correctness, edge cases, error handling, naming, and unclear intent. Severity levels — critical, warning, info.
Agent Claude Code
Adversarial review of a plan — finds gaps, optimistic assumptions, missing rollback paths, hand-waved steps, and hidden dependencies.
Agent Claude Code
Updates documentation to match code changes — README, CHANGELOG, inline docs, migration guides, examples.
Agent Claude Code
Validates that the development environment is correctly set up — dependencies, env vars, database, services, build tools.
Agent Claude Code
Frontend-specific reviewer for React/Vue/Angular — accessibility, component structure, hooks, render performance, state management.
Agent Claude Code
Triages production incidents. Stop the bleeding first, root-cause second, prevention third.
Agent Claude Code
Runs integration and end-to-end tests. Validates that components work together, not just individually.
Agent Claude Code
Resolves git merge conflicts by understanding both sides' intent. Combines compatible changes, escalates competing ones.
Agent Claude Code
Plans safe database migrations using the expand-backfill-contract pattern. Never destructive in production.
Agent Claude Code
Performance review — N+1 queries, unbounded fetches, missing pagination, sync work in async paths, heavy renders, allocations in tight loops.
Agent Claude Code
PHP/Laravel-specific reviewer — SQL injection, mass assignment, CSRF, N+1 Eloquent, auth middleware, type safety, queue idempotency.
Agent Claude Code
Turns architecture into a numbered, executable implementation plan with workstreams, file targets, and acceptance tests.
Agent Claude Code
Creates a rollback plan for any change — what to revert, in what order, and how to verify the rollback worked.
Agent Claude Code
OWASP Top 10 review on the diff. Flags injection, broken access control, crypto failures, SSRF, XSS, prompt injection. Adversarial mindset.
Agent Claude Code
Runs the project's test suite, captures results, identifies new failures, and flags missing coverage on changed code.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: