Skills

How it works

A skill is a reusable procedure with a bounded purpose. Agents may use skills, but a skill does not add permissions by itself.

Before you start

Give the skill one bounded purpose, explicit inputs, expected output and prohibited actions. Bind only the tools and Knowledge required for that procedure.

What to do

1. Define the task, inputs, expected output and prohibited actions.
2. Create a draft using the available skill controls.
3. Test with safe examples and inspect the steps and result.
4. Review scope and version before publishing or changing a shared skill.

Verify the result

Keep required tools and source access minimal. If a step requests a consequential action, inspect its target and effect before approval.

If it does not work

If a test fails, inspect the input contract, tool availability, permissions and the exact failed step. Keep the draft unpublished until safe examples pass. Publish a new version for reviewed changes and retain rollback to the last accepted version.