workflow-tools

Restore context after /clear - shows git state and active PEPs

Generate a headless claude command to autonomously implement a PEP wave. Usage - /run-pep [PEP file or number] [wave number]

You are a frontend testing specialist with browser access via claude-in-chrome.

You are a principal engineer evaluating whether a wave of work meets its acceptance criteria. You have read too many postmortems caused by "it looked done." You do not speculate. Every claim you make is backed by evidence you gathered yourself.

This skill should be used when the user asks to "clean up tests", "clean up code", "tidy up tests", "remove dead tests", "organize test files", "fix test structure", or mentions test hygiene, test bloat, removing temporary tests, or following test best practices. Unlike the code-review skill (which reviews a diff for quality), this skill performs batch cleanup across entire files or directories. Reports issues first, then applies fixes after confirmation.

This skill should be used when the user says "create a PEP", "update PEP", "PEP status", "document this", or references PEP-*.md files. Manages PEP (Project Enhancement Proposal) documents for tracking decisions and implementation plans.

This skill should be used when the user asks to "refactor this package", "reorganize this code", "split this file", "this code is a mess", "too many files", "simplify this module", "restructure this", or mentions god structs, god components, god classes, package organization, module structure, or code that is hard to navigate. Unlike the code-cleanup skill (which removes dead code and tests), this skill restructures working code for clarity and maintainability.

Reviews code changes for quality issues. Use when user says "review my changes", "review the diff", "review my code", "check for issues", "sanity check", or asks to find holes, gaps, or simplification opportunities.

This skill should be used when creating or editing documents, diagrams, handover docs, PEPs, or any substantial written output. Also activates when the user says "evaluate this", "run in a loop until good", "rate this out of 10", "is this 10/10", "loop until fixed", "keep improving", or asks for iterative quality improvement on any output.

1 hook across 1 event