| name | gpt2giga |
| description | Runs and configures the gpt2giga proxy that maps OpenAI API and Anthropic Messages API requests to GigaChat. Use when plugging GigaChat into existing OpenAI-compatible or Anthropic-compatible applications without rewriting the client integration. |
gpt2giga
Use this skill when the task is compatibility, migration, or proxy deployment.
What this skill covers
- local or Docker startup
.env and CLI configuration- OpenAI-compatible and Anthropic-compatible endpoints
- proxy auth and pass-through auth
- streaming and model routing behavior
- production hardening
Workflow
- Decide whether the user wants local development, Docker, or production deployment.
- Configure proxy auth separately from GigaChat auth.
- If the client must keep its OpenAI or Anthropic SDK, point it at the proxy base URL.
- If the client should carry its own GigaChat credentials, use pass-token mode.
- In production, enable proxy API-key auth and avoid DEV-only endpoints.
- When behavior differs from native OpenAI or Anthropic APIs, explain the mapping rather than hiding it.
- Prefer the smallest compatibility setup that preserves the client API the user already has.
Read these references as needed
- For env vars, ports, auth, and startup:
references/config.md
- For endpoint compatibility and client behavior:
references/compatibility.md
Default output
- treat
gpt2giga as a compatibility layer, not as the default for new native apps
- state which API family is being preserved: OpenAI or Anthropic
- call out that outer orchestration still owns full multi-step tool execution
Boundaries
- Do not teach raw SDK usage here; use
gigachat-sdk-*.
