skill-auto-improver

npx skills add luongnv89/asm --skill skill-auto-improver --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

# Skill Auto-Improver You run an eval-driven loop that **retrofits an existing SKILL.md to the current skill-creator standard**. This is the remediation tool for skills that did **not** go through skill-creator — external, legacy, manually-authored, or drifted. Fresh skill-creator output is publish-ready by construction (see skill-creator's `predictability-rubric.md` → _Publish-ready — no auto-improver dependency_) and should **not** normally need this skill. The target must clear **two hard gates**, then gets one **advisory** audit: 1. **Gate 1 — skill-creator standard (must-pass floor)** — `quick_validate` clean, Frontmatter Audit passes, ≤500 lines (detail below). 2. **Gate 2 — asm-eval floor (supplementary)** — `overallScore > 85` AND every category `>= 8`. 3. **Advisory — predictability audit (Phase 2b, not a gate)** — judgment-based findings against skill-creator's rubric, reported separately, **never** blocking. A skill that scores 92 but fails `quick_validate.py` is not done; one that passes `quick_validate.py` but scores 70 is not done. **Both gates must clear, or the loop reports a blocker** — open predictability findings alone never make one. ## Repo Sync Before Edits (m

git fetch origin
git pull --rebase origin "$branch"
test -f "$QV" || { echo "skill-creator not installed at $QV"; exit 1; }
test -f "$RUBRIC" || echo "⚠ predictability rubric missing — Phase 2b degraded (gates unaffected)"
mkdir -p .asm-improver
asm eval "$SKILL_PATH" --json > .asm-improver/baseline.json
python "$QV" "$SKILL_PATH" > .asm-improver/baseline-quickvalidate.txt 2>&1 || true
asm eval "$SKILL_PATH" --fix --dry-run   # preview the diff
asm eval "$SKILL_PATH" --fix              # write, creates SKILL.md.bak
asm eval "$SKILL_PATH" --json > .asm-improver/iter-1.json