TIANTIAN-ZHAN
UserScaffold a production AI agent harness in any repo, in 5 minutes. Tools for the post-vibe-coding era.
Categories
Indexed Skills (5)
bootstrap-harness
Create a complete AI-agent harness (CLAUDE.md + supporting docs system) for any project. Portable across Python/Java/TS/Go. Based on 2025-2026 best practices (AGENTS.md spec, GitHub's 2,500-repo study, ACE paper, context-engineering research). Use when a repo has no harness, was just cloned/scaffolded, or its docs are bloated/stale and need a structured rebuild.
finish-phase
Graduate a completed exec-plan from active/ to completed/, synthesize an outcome section (results / pitfalls / remaining), and update the index. Daily harness hygiene — keeps the project's "where are we now" pointer accurate so new agent sessions don't waste context rebuilding state.
promote-lesson
Find lessons in docs/lessons.md that have been referenced ≥2 times, then promote them into CLAUDE.md boundary rules via a single delta edit. Implements the ACE paper's structured-bullet + delta-edit pattern, so your harness evolves without full rewrites. The mechanism that makes the harness self-improving.
review-harness
Audit a project's harness (CLAUDE.md + supporting docs) against current code and git activity, and catch lifecycle/hygiene drift (completed work stuck in active/, stale or silently-parked plans, executed-runbook/dead-spec sections needing a tombstone, date-patch stacking, index↔directory drift, broken links or clickable links into a gitignored archive, status metadata contradicting the body, phantom-audience docs, double graveyards). Produces a markdown report only — never edits files. Use after a milestone, after a major line closeout, on a monthly cadence, or when the user runs /review-harness. Works in any repo with harness-shaped docs.
test-first
Test-driven development with red-green-refactor loop. Use when user wants to build features or fix bugs using TDD, mentions "red-green-refactor", wants integration tests, or asks for test-first development.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.