skill-catalog

name	skill-catalog
description	List the shipped video-evaluator skills so an agent can discover the shared review and artifact-intake surface.

Skill Catalog

Use this when an agent needs to discover the shipped capabilities in video-evaluator before choosing a narrower tool.

Inputs

Pass an empty JSON object:

{}

The runner reads JSON from stdin and returns JSON on stdout.

Outputs

The result lists each shipped skill with its name, description, example request path, and local runner hints. Treat the catalog as discovery metadata, not as a review result.

Sequencing Guidance

Run this first when you are dropped into an unfamiliar repo or an installed .video-evaluator pack and do not know which narrower skill is appropriate. After catalog discovery, choose the smallest skill that answers the task:

• Use video-artifact-intake to map a run directory before reviewing it.
• Use review-bundle for a fast status summary of one run.
• Use storyboard-extract, storyboard-ocr, storyboard-transitions, and storyboard-understand in that order for visual evidence.
• Use compare-video-runs only when two bundles are available.
• Use package-review-prompt when another agent needs a grounded prompt.

Abstention Rules

Do not run this repeatedly once you already know the needed skill. Do not use it as proof that artifacts exist; it only describes available capabilities.

Failure Modes

If the catalog runner is missing, the skill pack may not be installed correctly. Use the repo-side runner from the source checkout, or run install-skill-pack from the source repo if you need to materialize the pack elsewhere.

Repo-side runner:

node --import tsx scripts/harness/skill-catalog.ts

Installed-pack runner:

node ./.video-evaluator/agent/run-tool.mjs skill-catalog