skill-auto-improver

Name: skill-auto-improver
Author: luongnv89

742stars61forks

How to install skill-auto-improver?

luongnv89/asm/skill-auto-improver

$npx -y skills add luongnv89/asm --skill skill-auto-improver

Installs into the current project.

Summary

Improve an external, legacy, or drifted SKILL.md to the skill-creator standard — hard validation gates plus an advisory predictability audit. Don't use for authoring from scratch (skill-creator output is already standard), bulk eval, or prose edits.

Files · 1

View on GitHub

SKILL.md

1	# Skill Auto-Improver
2
3	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.
4
5	The target must clear two hard gates, then gets one advisory audit:
6
7	1. Gate 1 — skill-creator standard (must-pass floor) — `quick_validate` clean, Frontmatter Audit passes, ≤500 lines (detail below).
8	2. Gate 2 — asm-eval floor (supplementary) — `overallScore > 85` AND every category `>= 8`.
9	3. Advisory — predictability audit (Phase 2b, not a gate) — judgment-based findings against skill-creator's rubric, reported separately, never blocking.
10
11	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.
12
13	## Repo Sync Before Edits (mandatory)
14
15	This skill mutates files in a git repo. Before any edit, sync the local branch with the remote:
16
17	```bash
18	branch="$(git rev-parse --abbrev-ref HEAD)"
19	git fetch origin
20	git pull --rebase origin "$branch"
21	```
22
23	If the working tree is dirty, `git stash`, sync, then `git stash pop`. If `origin` is missing or `git pull` hits conflicts, stop and ask the user before continuing — do not skip or force the sync.
24
25	## When to Use
26
27	Reach for this on an existing, external, legacy, manually-authored, or drifted skill:
28
29	- The user asks to "improve", "level up", "fix", "polish", or "bring up to standard" an existing skill
30	- A skill was authored outside skill-creator (hand-written, imported, inherited) and must meet the current bar
31	- A skill has drifted — predates the standard, or edits left it failing `quick_validate.py` or below the 85/8 floor
32	- You are preparing such a skill for `asm publish` or a catalog
33
34	Not for routine cleanup of fresh skill-creator output (publish-ready by construction — use `/skill-creator` to author). Assumes a SKILL.md exists. For a report only, run `asm eval` and `quick_validate.py` directly — not this skill.
35
36	## Prerequisites
37
38	Verify all of the following before touching any files. Stop and tell the user if any fails.
39
40	- `asm` is available on PATH (`command -v asm` or `which asm`)
41	- Python 3 is available, and `~/.claude/skills/skill-creator/scripts/quick_validate.py` exists (skill-creator must be installed locally)
42	- The target skill path contains a `SKILL.md` file
43	- The working tree has no unrelated uncommitted edits (dirty files get mixed into diffs)
44	- You have write access to the skill directory
45
46	Resolve skill-creator's validator (required) and rubric (fail-soft) once and reuse them. The rubric is fail-soft — a locally-installed skill-creator may predate the repo and not ship it; a missing file only degrades Phase 2b to a warning and never aborts the run:
47
48	```bash
49	QV="$HOME/.claude/skills/skill-creator/scripts/quick_validate.py"
50	test -f "$QV" \|\| { echo "skill-creator not installed at $QV"; exit 1; }
51	RUBRIC="$HOME/.claude/skills/skill-creator/references/predictability-rubric.md"
52	test -f "$RUBRIC" \|\| echo "⚠ predictability rubric missing — Phase 2b degraded (gates unaffected)"
53	```
54
55	## Inputs
56
57	The user provides one of:
58
59	- A local skill path: `skills/foo` or `/abs/path/to/skill`
60	- A direct `SKILL.md` file path (treated as its parent directory)
61	- A GitHub shorthand: `github:owner/repo` or `github:owner/repo:path/to/skill`
62
63	For GitHub inputs, ask the user to clone locally first or whether you should open a PR back to that repo. This skill's default path is local editing — remote editing is out of scope for v1.
64
65	## The Gates
66
67	Keep two decision classes separate: hard gates (Gate 1, Gate 2) are mechanical and pass/fail — they alone decide PASS vs BLOCKER. Predictability findings (Phase 2b) are judgment-based and advisory — both gates green with open findings is still a PASS.
68
69	### Gate 1 — Skill-creator standard (must-pass floor)
70
71	A skill passes this gate when all of these are true:
72
73	- `python "$QV" "$SKILL_PATH"` exits 0 (no unexpected keys, name is kebab-case ≤64 chars, description is single-line ≤1024 chars, etc.)
74	- The Frontmatter Audit (full checklist in `refere

People who liked this also viewed

6 picks

Type

skill-auto-improver

742stars61forks

View on GitHub

How to install skill-auto-improver?

luongnv89/asm/skill-auto-improver

$npx -y skills add luongnv89/asm --skill skill-auto-improver

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Skill Auto-Improver
2
3	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.
4
5	The target must clear two hard gates, then gets one advisory audit:
6
7	1. Gate 1 — skill-creator standard (must-pass floor) — `quick_validate` clean, Frontmatter Audit passes, ≤500 lines (detail below).
8	2. Gate 2 — asm-eval floor (supplementary) — `overallScore > 85` AND every category `>= 8`.
9	3. Advisory — predictability audit (Phase 2b, not a gate) — judgment-based findings against skill-creator's rubric, reported separately, never blocking.
10
11	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.
12
13	## Repo Sync Before Edits (mandatory)
14
15	This skill mutates files in a git repo. Before any edit, sync the local branch with the remote:
16
17	```bash
18	branch="$(git rev-parse --abbrev-ref HEAD)"
19	git fetch origin
20	git pull --rebase origin "$branch"
21	```
22
23	If the working tree is dirty, `git stash`, sync, then `git stash pop`. If `origin` is missing or `git pull` hits conflicts, stop and ask the user before continuing — do not skip or force the sync.
24
25	## When to Use
26
27	Reach for this on an existing, external, legacy, manually-authored, or drifted skill:
28
29	- The user asks to "improve", "level up", "fix", "polish", or "bring up to standard" an existing skill
30	- A skill was authored outside skill-creator (hand-written, imported, inherited) and must meet the current bar
31	- A skill has drifted — predates the standard, or edits left it failing `quick_validate.py` or below the 85/8 floor
32	- You are preparing such a skill for `asm publish` or a catalog
33
34	Not for routine cleanup of fresh skill-creator output (publish-ready by construction — use `/skill-creator` to author). Assumes a SKILL.md exists. For a report only, run `asm eval` and `quick_validate.py` directly — not this skill.
35
36	## Prerequisites
37
38	Verify all of the following before touching any files. Stop and tell the user if any fails.
39
40	- `asm` is available on PATH (`command -v asm` or `which asm`)
41	- Python 3 is available, and `~/.claude/skills/skill-creator/scripts/quick_validate.py` exists (skill-creator must be installed locally)
42	- The target skill path contains a `SKILL.md` file
43	- The working tree has no unrelated uncommitted edits (dirty files get mixed into diffs)
44	- You have write access to the skill directory
45
46	Resolve skill-creator's validator (required) and rubric (fail-soft) once and reuse them. The rubric is fail-soft — a locally-installed skill-creator may predate the repo and not ship it; a missing file only degrades Phase 2b to a warning and never aborts the run:
47
48	```bash
49	QV="$HOME/.claude/skills/skill-creator/scripts/quick_validate.py"
50	test -f "$QV" \|\| { echo "skill-creator not installed at $QV"; exit 1; }
51	RUBRIC="$HOME/.claude/skills/skill-creator/references/predictability-rubric.md"
52	test -f "$RUBRIC" \|\| echo "⚠ predictability rubric missing — Phase 2b degraded (gates unaffected)"
53	```
54
55	## Inputs
56
57	The user provides one of:
58
59	- A local skill path: `skills/foo` or `/abs/path/to/skill`
60	- A direct `SKILL.md` file path (treated as its parent directory)
61	- A GitHub shorthand: `github:owner/repo` or `github:owner/repo:path/to/skill`
62
63	For GitHub inputs, ask the user to clone locally first or whether you should open a PR back to that repo. This skill's default path is local editing — remote editing is out of scope for v1.
64
65	## The Gates
66
67	Keep two decision classes separate: hard gates (Gate 1, Gate 2) are mechanical and pass/fail — they alone decide PASS vs BLOCKER. Predictability findings (Phase 2b) are judgment-based and advisory — both gates green with open findings is still a PASS.
68
69	### Gate 1 — Skill-creator standard (must-pass floor)
70
71	A skill passes this gate when all of these are true:
72
73	- `python "$QV" "$SKILL_PATH"` exits 0 (no unexpected keys, name is kebab-case ≤64 chars, description is single-line ≤1024 chars, etc.)
74	- The Frontmatter Audit (full checklist in `refere

People who liked this also viewed

6 picks

Type