agentsclimarketplace

Openmind capability router

Skill HelloThisWorld/agent-skill-verification-template/skills/openmind-capability-router

Production-oriented template for building AI agent skills as verifiable software components — offline eval harness, source-grounding validators, structured logs/traces/metrics, replay artifacts, and a CI quality gate. Runs fully offline with a deterministic mock model.

Install
npx -y skills add HelloThisWorld/agent-skill-verification-template --skill openmind-capability-router

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Route any query to exactly one documented Open Mind capability (glossary / structure / search) using the deterministic if-else floor — the mode where routing never depends on a model. The decision is grounded by citing the capability's documented registry line. Runs Open Mind's real Python implementation through the skill bridge.

SKILL.md

1.4 KB, as published. Nobody here has run it

openmind-capability-router

This skill is the eval-harness face of Open Mind's capability-router (https://github.com/HelloThisWorld/open-mind, skills/capability-router/SKILL.md). Decisions come from Open Mind's actual router (openmind/router.py::route) in use_model=False mode — the deterministic floor an offline, reproducible eval can meaningfully gate on. In the app, a ready local model MAY refine the choice, but its pick is validated against the same closed capability set; this eval verifies the floor that behavior degrades to.

Behavior under contract

  • Closed capability set — every query maps to glossary, structure, or search; there is no path that yields anything else.
  • Full trace — the answer carries capability, decided_by, deterministic_fallback, and reason, so routing is auditable.
  • Grounded decision — the claim cites the registry line documenting the chosen capability; an invented capability would have no citable line and fail validation.

Question form

Any query string; it is routed as-is.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.