# openrouter-secrets

An OpenAI-compatible chat endpoint for DeepSeek V4.1 Flash on OpenRouter. The
capsule only forwards requests; m64's gate adds the OpenRouter key on the way
out, so the capsule never holds it.

Base URL: `{{BASE_URL}}`

## Routes

| Method | Path | What |
| --- | --- | --- |
| POST | `/v1/chat/completions` | chat completions, streaming or not |
| GET | `/v1/models` | one model: `deepseek-v4.1-flash` |
| GET | `/status` | phase (`ready` when usable) |
| GET | `/human/` | a page for people |

```sh
curl -s {{BASE_URL}}/v1/chat/completions {{AUTH}} \
  -H 'Content-Type: application/json' \
  -d '{"messages":[{"role":"user","content":"Say hi"}],"max_tokens":50}'

curl -N {{BASE_URL}}/v1/chat/completions {{AUTH}} \
  -H 'Content-Type: application/json' \
  -d '{"messages":[{"role":"user","content":"Count to 5"}],"stream":true}'
```

Accepted fields: `messages` (required), `tools`, `tool_choice`,
`response_format`, `max_tokens`, `stream`, `seed`. Others, like `temperature`,
get a 400. The model is always DeepSeek V4.1 Flash, whatever `model` says.

## Timings and errors

- Ready in a few seconds. Answers take as long as OpenRouter takes.
- OpenRouter's errors keep their status and body: 401 means the key is missing
  or wrong, 429 a rate limit.
- 502 means openrouter.ai could not be reached: the capsule was served without
  `--allow-external-network`.
- Requests are handled in parallel. Every request spends credit on the key's
  OpenRouter account.