For the complete documentation index, see llms.txt. This page is also available as Markdown.

Skill Library

Every first-party skill, one page each.

A skill is a single action an agent can take. This section documents every first-party skill the platform ships — what it does, a worked example, its configuration, its output, and how it fails.

For what skills are and how the planner chooses between them, see Skills & Tools. To attach one to an agent, see Adding a Skill to the Agent.

By category

Reading these pages

Each page follows the same spine: purpose → how it works → worked example → configuration → output → errors. Configuration and output tables are the contract; the worked example shows one real use, not the only one.

Where a skill supports it, a Testing in isolation section gives the endpoint for running that skill alone — useful for confirming a configuration before wiring it into anything.

Five skills are missing their reference layer. Intent Classification, Prompt, RAG Query, Ruleset and Vector Search document their purpose and builder configuration, but their formal schemas and error codes are not yet available. Those pages say so at the top.

Last updated