Skill Validator (Scenario-specific)

An internal skill that spins up a fresh, isolated agent to stress-test a published Scenario skill and reports every documentation defect back on the PR.

Dev & CodingAdvanced★ 630⑂ 76AI score 7/10Last updated: Sep 26, 2026

What it does

Reads the target skill's SKILL.md and linked files, then restates the objective, the claimed triggers, and the three to six non-obvious facts ("traps") the skill exists to teach.
Runs the free mechanical checks first: pnpm spec, plus pnpm test when the skill ships a script.
Builds a run directory outside the repo, installs the working-tree version of the skill, and launches a genuinely fresh agent via the Claude Code CLI (or a subagent fallback) to do real work with a budget and explicit success criteria.
Grades the run against a strict rubric: objective met, real tool names only, correct loop, no hardcoded generative model ids, MCP-only for the agent's own steps, traps handled, no guessing.
Emits pass / pass with notes / fail, ties each defect to the exact SKILL.md line to change, and posts the report as a PR comment with redaction rules for asset URLs, team and project.

Who it's for

Maintainers contributing to or reviewing skills in the scenario-labs/skills repository.
Users who already have a connected Scenario MCP server, generation credits, pnpm, gh, and the Claude Code CLI.
Note: it is flagged internal: true and disable-model-invocation: true, so it is effectively unusable outside that repo.

Examples

/skills:validate scenario-comfyui — a live run with one minimal generation to prove the upload flow and jobs_wait re-calls are taught correctly.
/skills:validate scenario --plan-only — zero-cost grading of a numbered tool-call plan.
/skills:validate scenario-blender --pr 412 --no-post — validate against a specific PR context but keep the report in the session.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/scenario-labs/skills/HEAD/.agents/skills/skills-validate/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)

Download the ZIP with the button below.
In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.

↓ Download ZIP

Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the .agents/skills/skills-validate folder from the GitHub repo scenario-labs/skills into my ~/.claude/skills/skills-validate/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/scenario-labs/skills.git && mkdir -p ~/.claude/skills && cp -R skills/.agents/skills/skills-validate ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

Open a terminal in the folder where you keep code.
Clone the repository: git clone https://github.com/scenario-labs/skills.git
Create the skills folder: mkdir -p ~/.claude/skills
Copy the skill: cp -R skills/.agents/skills/skills-validate ~/.claude/skills/
Install the prerequisites: Node.js with pnpm, GitHub CLI (gh), the Claude Code CLI, and a connected Scenario MCP server.
Launch Claude Code from the root of the scenario-labs/skills repo, since the skill references scripts/lib/skills.mjs, AGENTS.md, and .claude/agents/skill-tester.md.
Invoke it as /skills:validate <skill-name>; start with --plan-only to avoid spending Scenario credits.

View source on GitHub ↗License: MIT