Agent Automation & Research Skills

Sweep a codebase for AI-accumulated cruft (dead code, stale comments, compatibility shims, defensive bloat, over-abstraction), triage findings by confidence tier, then orchestrate parallel cleanup agents with worktree isolation. Use when user says "clean cruft", "dead code", "unused code", "stale comments", or "AI leftovers".

ALWAYS invoke when the user asks to implement substrate items, work through stage:implementing items, drain the queue, or implement a feature/epic scope. Default implementation path; call implement directly only when the user says "inline". Builds a dependson graph, bundles related work, chooses wave width and worktree isolation, and dispatches implementation sub-agents when useful. This skill authorizes sub-agents, including large non-overlapping write paths when ownership and verification make that safe. Advances parents whose children all reach stage:review.

Cut a release in the agile-workflow substrate. Interactively binds items to a version, advances the release stage from planned to quality-gate, runs all configured gates in CONVENTIONS.md order (default: security, tests, cruft, docs, patterns), waits until all bound items + gate-produced items reach stage:done, ships per release mapping (tag-based / branch-held / release-branch / none), collapses bound items into one release summary and disposes their bodies per the terminal-tier retention convention (delete-refs prunes to git history; retain-bodies keeps bodies on disk), advances release to released. Idempotent — safe to re-run after fixing gate findings.

Multi-dimensional codebase evaluation with verified scoring. Launches parallel explore agents, cross-checks findings with direct verification, produces calibrated 1-10 scorecard across architecture, code quality, testing, documentation, CI/CD, error handling, security, DX, and maintainability with prioritized recommendations. Use when user asks to evaluate, audit, score, or review a repository holistically.

Z.ai-powered web research for Pi: web search, URL/PDF/JSON/API reading, and GitHub repo deep-read. Reaches for five agent tools — zaiwebsearch (Z.ai web search), fetchcontent (webReader pages, local PDF extraction, direct JSON/API fetches, and article extraction for noisy docs), searchrepodocs / getrepostructure / readrepofile (Z.ai zread over GitHub). Use whenever work needs current or version-sensitive information, structured API metadata, or deep-reading a GitHub repo without cloning. Auth comes from the configured Z.ai (zai) provider key.