Files

T

infosave2007 cbc385f6c7 feat: universal LLM gateway with embedded multilingual admin console

- OpenAI-compatible gateway with intelligent routing via cortiq-router (allaigate)
- Embedded build-free admin SPA at /admin: 7 languages (en/ru/de/fr/es/zh/tr),
  light/dark themes, hot config reload (no restart)
- Admin API, per-request statistics + Prometheus /metrics, secret store, playground
- English documentation (primary) with Russian translations (*.ru.md)

2026-06-30 10:51:37 +03:00

1.7 KiB

Raw Blame History

Read this in Русский.

Contributing to Cortiq Gateway

Thank you for your interest in the project! This is an open gateway and we welcome contributions.

Getting started

Read docs/ARCHITECTURE.md — design overview and roadmap.
Review docs/PROTOCOLS.md and docs/ROUTING.md.
The current phase is Phase 0 (skeleton). Good entry-point tasks are labeled good first issue.

How to add...

A new inbound protocol — implement the InboundProtocol trait in src/protocols/, translate to/from the canonical model (ChatRequest/ChatResponse), and add the corresponding flag in [protocols]. Do not touch the providers.
A new provider — implement the Provider trait in src/providers/. Most local servers are OpenAI-compatible, so reusing openai.rs is often sufficient.
A new routing policy — extend src/routing.rs and the [routing.policy] section.

Core principle: N protocols × M providers communicate only through the canonical model — no direct protocol↔provider translations.

Style and checks

cargo fmt --all
cargo clippy --all-targets -- -D warnings
cargo test

Match the style of the surrounding code.
Secrets go only through *_env variables — never commit them to the repository.
New functionality must be accompanied by tests and updates to docs/.

Pull requests

Keep PRs small and focused. One PR = one logical unit of change.
In the description: what, why, and how it was tested.
Link the related issue if one exists.

License

Contributions are accepted under Apache-2.0.

1.7 KiB Raw Blame History Unescape Escape