review-skill

Name: review-skill
Author: agent-ecosystem

191stars29forks

How to install review-skill?

agent-ecosystem/skill-validator/review-skill

$npx -y skills add agent-ecosystem/skill-validator --skill review-skill

Installs into the current project.

Summary

Review a proposed Agent Skill for structural validity and content quality before publishing. Runs the skill-validator CLI to check for structural issues, scores the skill with an LLM judge, and interprets results to advise authors on what to address. Use when a user wants to revi

Files · 1

View on GitHub

SKILL.md

1	# Review Skill Workflow
2
3	You are helping a skill author review an Agent Skill before publishing. This is
4	a multi-step process: determine environment, verify prerequisites, run
5	structural validation, review content, optionally run LLM scoring, and
6	interpret results. Follow every step in order.
7
8	## Step 0: Determine Environment
9
10	Check for saved configuration:
11
12	```bash
13	cat ~/.config/skill-validator/review-state.yaml 2>/dev/null
14	```
15
16	If the state file exists with `prereqs_passed: true`, offer:
17
18	> Found saved settings — configured for [provider/structural-only] reviews.
19	>
20	> 1. Continue with saved settings — skip to Step 2
21	> 2. Re-run prerequisite checks
22	> 3. Change environment — switch provider or between LLM and structural-only
23
24	Option 1: read `llm_scoring`, `provider`, and `cross_model` from the file and
25	skip to Step 2.
26	Options 2-3: continue below.
27
28	If no state file exists, or the user chose to re-check/change, ask:
29
30	> LLM scoring uses an Anthropic or OpenAI-compatible API, or the Claude CLI.
31	> Without an API key or CLI, we run structural validation only.
32	>
33	> 1. Anthropic — use Claude via the Anthropic API (requires `ANTHROPIC_API_KEY`)
34	> 2. OpenAI — use GPT via the OpenAI API (requires `OPENAI_API_KEY`)
35	> 3. OpenAI-compatible — use a custom endpoint (Ollama, Groq, Azure, Together, etc.)
36	> 4. Claude CLI — use the locally authenticated `claude` binary (no API key needed)
37	> 5. Skip LLM scoring — structural validation only
38
39	Options 1-4: set `LLM_SCORING=true` and record the provider choice.
40	Option 5: set `LLM_SCORING=false`. Run Step 1a only, then jump to Step 2.
41
42	If the user chose option 1 or 2, ask about cross-model comparison:
43
44	> Scoring with a second model family gives more robust novelty scores, since
45	> each model has different training data. This requires API keys for both
46	> Anthropic and OpenAI.
47	>
48	> 1. Yes, compare across model families — score with both Anthropic and OpenAI
49	> 2. No, single provider is fine
50
51	Option 1: set `CROSS_MODEL=true`. Option 2: set `CROSS_MODEL=false`.
52
53	Do not offer cross-model comparison for option 3 (OpenAI-compatible) or option 4
54	(Claude CLI), since the second provider would need a standard Anthropic or
55	OpenAI key.
56
57	After Step 1a, follow
58	[references/llm-scoring.md](references/llm-scoring.md) for API key checks
59	before Step 2.
60
61	## Step 1: Verify Prerequisites
62
63	### 1a. Check for `skill-validator` binary
64
65	```bash
66	skill-validator --version
67	```
68
69	If not found, search common locations (`/usr/local/bin`, `/opt/homebrew/bin`,
70	`~/go/bin`). If found but not on PATH, tell the user. If not found anywhere,
71	follow [references/install-skill-validator.md](references/install-skill-validator.md).
72
73	Do NOT proceed until this succeeds.
74
75	If `LLM_SCORING=true`, complete the provider checks in
76	[references/llm-scoring.md](references/llm-scoring.md) before continuing.
77
78	### Save state after prerequisites pass
79
80	Persist state so future runs skip this step. Replace placeholders with actual
81	values:
82
83	```bash
84	mkdir -p ~/.config/skill-validator
85	cat > ~/.config/skill-validator/review-state.yaml << 'EOF'
86	prereqs_passed: true
87	llm_scoring: <true or false>
88	provider: <anthropic, openai, openai-compatible, or claude-cli>
89	model: <model name if specified, or "default">
90	base_url: <custom base URL if openai-compatible, or omit>
91	cross_model: <true or false>
92	EOF
93	```
94
95	## Step 2: Locate the Skill
96
97	Ask the user for the path to the skill they want to review, unless they have
98	already provided it. Verify the path contains a `SKILL.md` file:
99
100	```bash
101	ls <path>/SKILL.md
102	```
103
104	If `SKILL.md` does not exist at the given path, tell the user this is not a
105	valid skill directory and ask them to provide the correct path.
106
107	## Step 3: Run Structural Validation
108
109	Run the full check suite:
110
111	```bash
112	skill-validator check <path>
113	```
114
115	Capture the exit code:
116
117	\| Exit code \| Meaning \|
118	\|-----------\|---------\|
119	\| 0 \| Clean — no errors or warnings \|
120	\| 1 \| Errors found — must fix before publishing \|
121	\| 2 \| Warnings only — review but not blocking \|
122	\| 3 \| CLI/usage error — check the command \|
123
124	Exit 0: proceed. Exit 2: note warnings, proceed. Exit 1: list errors — these
125	are blocking. The user must fix them before the skill can be published. Do NOT
126	proceed to LLM scoring if exit code is 1.
127
128	## Step 4: Content Review
129
130	Read the SKILL.md and any reference files, then evaluate each check below.
131	Report which checks pass and which do

People who liked this also viewed

6 picks

Type

review-skill

191stars29forks

View on GitHub

How to install review-skill?

agent-ecosystem/skill-validator/review-skill

$npx -y skills add agent-ecosystem/skill-validator --skill review-skill

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Review Skill Workflow
2
3	You are helping a skill author review an Agent Skill before publishing. This is
4	a multi-step process: determine environment, verify prerequisites, run
5	structural validation, review content, optionally run LLM scoring, and
6	interpret results. Follow every step in order.
7
8	## Step 0: Determine Environment
9
10	Check for saved configuration:
11
12	```bash
13	cat ~/.config/skill-validator/review-state.yaml 2>/dev/null
14	```
15
16	If the state file exists with `prereqs_passed: true`, offer:
17
18	> Found saved settings — configured for [provider/structural-only] reviews.
19	>
20	> 1. Continue with saved settings — skip to Step 2
21	> 2. Re-run prerequisite checks
22	> 3. Change environment — switch provider or between LLM and structural-only
23
24	Option 1: read `llm_scoring`, `provider`, and `cross_model` from the file and
25	skip to Step 2.
26	Options 2-3: continue below.
27
28	If no state file exists, or the user chose to re-check/change, ask:
29
30	> LLM scoring uses an Anthropic or OpenAI-compatible API, or the Claude CLI.
31	> Without an API key or CLI, we run structural validation only.
32	>
33	> 1. Anthropic — use Claude via the Anthropic API (requires `ANTHROPIC_API_KEY`)
34	> 2. OpenAI — use GPT via the OpenAI API (requires `OPENAI_API_KEY`)
35	> 3. OpenAI-compatible — use a custom endpoint (Ollama, Groq, Azure, Together, etc.)
36	> 4. Claude CLI — use the locally authenticated `claude` binary (no API key needed)
37	> 5. Skip LLM scoring — structural validation only
38
39	Options 1-4: set `LLM_SCORING=true` and record the provider choice.
40	Option 5: set `LLM_SCORING=false`. Run Step 1a only, then jump to Step 2.
41
42	If the user chose option 1 or 2, ask about cross-model comparison:
43
44	> Scoring with a second model family gives more robust novelty scores, since
45	> each model has different training data. This requires API keys for both
46	> Anthropic and OpenAI.
47	>
48	> 1. Yes, compare across model families — score with both Anthropic and OpenAI
49	> 2. No, single provider is fine
50
51	Option 1: set `CROSS_MODEL=true`. Option 2: set `CROSS_MODEL=false`.
52
53	Do not offer cross-model comparison for option 3 (OpenAI-compatible) or option 4
54	(Claude CLI), since the second provider would need a standard Anthropic or
55	OpenAI key.
56
57	After Step 1a, follow
58	[references/llm-scoring.md](references/llm-scoring.md) for API key checks
59	before Step 2.
60
61	## Step 1: Verify Prerequisites
62
63	### 1a. Check for `skill-validator` binary
64
65	```bash
66	skill-validator --version
67	```
68
69	If not found, search common locations (`/usr/local/bin`, `/opt/homebrew/bin`,
70	`~/go/bin`). If found but not on PATH, tell the user. If not found anywhere,
71	follow [references/install-skill-validator.md](references/install-skill-validator.md).
72
73	Do NOT proceed until this succeeds.
74
75	If `LLM_SCORING=true`, complete the provider checks in
76	[references/llm-scoring.md](references/llm-scoring.md) before continuing.
77
78	### Save state after prerequisites pass
79
80	Persist state so future runs skip this step. Replace placeholders with actual
81	values:
82
83	```bash
84	mkdir -p ~/.config/skill-validator
85	cat > ~/.config/skill-validator/review-state.yaml << 'EOF'
86	prereqs_passed: true
87	llm_scoring: <true or false>
88	provider: <anthropic, openai, openai-compatible, or claude-cli>
89	model: <model name if specified, or "default">
90	base_url: <custom base URL if openai-compatible, or omit>
91	cross_model: <true or false>
92	EOF
93	```
94
95	## Step 2: Locate the Skill
96
97	Ask the user for the path to the skill they want to review, unless they have
98	already provided it. Verify the path contains a `SKILL.md` file:
99
100	```bash
101	ls <path>/SKILL.md
102	```
103
104	If `SKILL.md` does not exist at the given path, tell the user this is not a
105	valid skill directory and ask them to provide the correct path.
106
107	## Step 3: Run Structural Validation
108
109	Run the full check suite:
110
111	```bash
112	skill-validator check <path>
113	```
114
115	Capture the exit code:
116
117	\| Exit code \| Meaning \|
118	\|-----------\|---------\|
119	\| 0 \| Clean — no errors or warnings \|
120	\| 1 \| Errors found — must fix before publishing \|
121	\| 2 \| Warnings only — review but not blocking \|
122	\| 3 \| CLI/usage error — check the command \|
123
124	Exit 0: proceed. Exit 2: note warnings, proceed. Exit 1: list errors — these
125	are blocking. The user must fix them before the skill can be published. Do NOT
126	proceed to LLM scoring if exit code is 1.
127
128	## Step 4: Content Review
129
130	Read the SKILL.md and any reference files, then evaluate each check below.
131	Report which checks pass and which do

People who liked this also viewed

6 picks

Type