README for the Inference Gateway — per PRD §4 the security boundary, the only component holding privileged provider keys and the only egress path for customer prompts. Exposes an OpenAI-compatible surface (/v1/chat/completions, /v1/embeddings) plus admin endpoints under /admin/v1, and reads a single gateway.yaml config (schema in gateway.yaml.example). Same run/test shape as the api README and the same stale "M1 build in progress" status despite shipped M2 features. You need the gateway's run/test commands, its endpoint paths, or where its config schema lives.
LQ.AI Inference Gateway
Per PRD §4, the gateway is the security boundary — the only component holding privileged provider API keys, and the only egress path for customer prompts.
OpenAI-compatible surface (/v1/chat/completions, /v1/embeddings) plus admin endpoints under /admin/v1.
Quick start
python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
uvicorn app.main:app --reload --port 8001
Health check: curl http://localhost:8001/health
Configuration
The gateway reads a single YAML config (gateway.yaml); see ../gateway.yaml.example for the full schema.
Tests
pytest # unit + integration
pytest -m "not provider" # skip real-provider integration tests
ruff check .
ruff format --check .
mypy .
Status
M1 build in progress. See ../docs/M1-IMPLEMENTATION-ORDER.md for task breakdown.