generated: '2026-07-21' method: searched source: https://github.com/shapesinc/shapes-api (README) + openapi/shapes-api-openapi.yml summary: >- Cross-cutting semantics for the Shapes API, an OpenAI-compatible chat API. Grounded in the official repository documentation. authentication: style: bearer header: 'Authorization: Bearer ' env: SHAPESINC_API_KEY see: authentication/shapes-authentication.yml compatibility: standard: openai detail: >- Follows the OpenAI chat completions request/response contract; works with existing OpenAI SDKs by setting base_url to https://api.shapes.inc/v1/. model_addressing: scheme: shapesinc/ detail: The `model` field names the Shape to run rather than a raw model id. identity_headers: - name: X-User-Id purpose: Per-user identification / memory + context isolation. - name: X-Channel-Id purpose: Per-conversation context for multi-conversation apps. request_id_tracing: supported: unknown detail: Not documented. idempotency: supported: false detail: No idempotency keys or headers are documented. pagination: supported: false detail: No list pagination is documented for the public endpoints. versioning: style: uri detail: Version segment `/v1/` in the base URL. see: lifecycle/shapes-lifecycle.yml rate_limiting: default: 20 requests per minute increase: https://docs.google.com/forms/d/e/1FAIpQLScGLeRk6snViRPslXbbUaMDwubcBhmcJ6opq7wFvPEp-EbO3g/viewform signaling: Not documented (no documented rate-limit response headers). error_envelope: shape: openai-compatible problem_json: false detail: Errors returned in standard OpenAI-compatible JSON; RFC 9457 problem+json is not used. multimodal: detail: One image_url OR one audio_url per user message; if both are sent only the audio is processed. audio_formats: [mp3, wav, ogg] limitations: - No system messages (personality comes from Shape configuration). - No message history passed by client (relies on Shape memory). - No streaming responses. - No client parameter control (temperature etc. locked to Shape settings). commands: - '!reset - reset long-term memory' - '!wack - reset short-term memory' - '!sleep - generate long-term memory on demand' - '!web - search the web' - '!imagine - generate images' - '!info / !help / !dashboard'