//
Search across all documentation pages
8 pages in this section.
Explains the properties (idempotency, clear errors, scoped access) that make an API safe to expose as an agent tool.
A first custom tool wrapping a simple internal REST endpoint for agent function calling.
A recipe for turning an existing REST endpoint into a callable agent tool with schema, auth, and error mapping.
A recipe for giving an agent a scoped web-search tool with source citations and safe result shaping.
A checklist for evolving tool schemas without silently breaking agents that call them.
Ten practices for designing internal tools that agents call safely and correctly.