This documentation is the product. There is no playground. There is no chat UI. If you want to evaluate HermesBridge, run the quickstart against a sandbox identity.
POST /v1/chat/completions
Primary completion endpoint. OpenAI-compatible request shape with two additions: the X-Hermes-Signature header and an extended usage block in the response.
Request
Request body fields
| field | type | description |
|---|---|---|
| model | string | Model ID, or a routing mode keyword (auto, cheapest, fastest, balanced, capability:<x>). |
| messages | array | OpenAI-compatible message array. Standard roles: system, user, assistant, tool. |
| max_tokens | integer | Maximum tokens in the completion. Optional. |
| temperature | float | Sampling temperature 0–2. Optional, default 1.0. |
| routing | object | HermesBridge routing constraints. Optional. See routing fields below. |
| stream | boolean | Stream the response via SSE. Optional, default false. |
Response
The response follows the OpenAI chat completion shape. The usage block includes HermesBridge-specific fields: agent_did, attestation_tier, cost_usd, discount_applied, and cache_hit.
Error response
See Error taxonomy for all error codes.