Atuin AI Proxy lets developers route shell AI through any OpenAI-compatible backend

Atuin AI Proxy addresses a common constraint in developer tooling: the need to integrate with AI services while maintaining flexibility. By inserting a translation layer between Atuin's native API and OpenAI-compatible backends, the tool lets users avoid coupling to a single AI vendor. This is particularly useful for organizations that want to experiment with different models, optimize costs, or comply with data residency requirements.

The proxy's design prioritizes observability and ease of troubleshooting. It assigns a request ID to every HTTP response and includes it in stream errors, allowing developers to correlate Atuin's SSE (Server-Sent Events) failures with upstream backend issues. The logging system offers granular control: INFO for normal operation, DEBUG for model and timing diagnostics, TRACE for sanitized payloads (capped at configurable byte limits to avoid exposing sensitive shell history). This layered approach supports both production monitoring and development diagnosis without forcing operators to log sensitive data by default.

Deployment is simplified by Docker Compose and local development is minimal — the proxy uses only Python's standard library at runtime, reducing dependency management and container bloat. The tool handles the semantic differences between Chat Completions and Responses APIs automatically: in 'auto' mode, it tries Chat Completions first and falls back to Responses if the upstream service rejects it as unsupported.