Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.
Audit test coverage for a stage or file by reading branches, identify gaps, write missing tests, and verify all tests pass. Use this skill whenever the user wants to add tests, check coverage, fill test gaps, or ensure a file is properly tested. Triggers on: "write tests", "add tests", "fill test gaps", "check coverage", "make sure this is tested", "tests are missing", or any request to improve test coverage for a specific file or stage.
Update documentation in this repository using Markdown. Use when asked to add, revise, or verify docs content in docs/, README.md, AGENTS.md, or any documentation file; includes edit workflow, build validation, and change summary expectations.
Run Python quality gates in order after code edits: format, lint, and type-check, and resolve all remaining issues before finishing. Use this skill after any edit to .py files, before committing, or whenever ruff or ty errors are reported. Triggers on: "fix quality", "run quality gates", "lint errors", "type errors", "ruff check", "ty check", "clean up the code", or any request to verify code quality after making changes.
Audit and fix logging in a file or stage so every significant event is logged at the right level — no silent failures, no noise, no missing context. Use this skill after an audit flags silent error paths or missing log coverage, when a bug report comes in and the log file does not explain what happened, or before releasing a stage. Triggers on: "fix logging", "add logging", "missing logs", "silent failure", "log coverage", "improve logging", or any request to ensure a file or stage is properly instrumented.
Act on audit findings: remove dead code, eliminate unnecessary wrappers and flags, collapse duplication, rename for clarity, and refactor for simplicity. Runs quality gates, fixes logging, and updates docstrings after every change. Use this skill whenever the user wants to clean up, simplify, or improve code structure after an audit. Triggers on: "fix the code", "clean this up", "remove dead code", "simplify this", "act on the audit findings", "refactor", or any request to improve code quality based on findings.
Systematically examine a file or stage: read everything in scope, challenge every line, and produce a ranked findings report — no code changes. Use this skill whenever the user wants to understand, review, or assess code before touching it. Triggers on: "audit", "deep dive", "review", "scrutinise", "look at", "go through", "what is this doing", "is this good", "check this stage", or any request to understand code before making changes.