subagentproducts

Direct developer access to Claude via a documented HTTP API (the Anthropic Messages API and related endpoints), with no built-in chat UI of its own — the surface through which every other product in this catalog, and countless third-party applications, ultimately reach a Claude model. Includes developer-facing platform features like prompt caching, tool use, and (per this repo's own docs mirror) Managed Agents primitives (sessions, memory stores, dreams, permission policies).

The Platform/API is the most "model-adjacent" surface in this catalog — a thin, direct pass-through to a chosen model — but it is still a surface, not a model itself: the same API endpoint can be pointed at different model IDs (claude-sonnet-5, claude-fable-5, etc.) via a request parameter.

General knowledge as of this session, and directly corroborated by this repo's own docs mirror under docs/docs/platform.claude.com/ (citations.md, prompt-caching.md, and the whole managed-agents/ subtree already cited by sibling primitives in this repo).

created 2026-07-02 08:28:37 · JSON