Skip to content
View happy520ai's full-sized avatar

Block or report happy520ai

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
happy520ai/README.md

Jiongjiong Chen | Open-Source AI Infrastructure

I build terminal-first infrastructure for models, agents, knowledge, and tools, with explicit human control at every real-provider boundary.

陈炯炯,正在构建开放、终端优先、可治理,并始终由人掌握最终控制权的 AI 基础设施。

Featured Project: Unified AI System

Unified AI System is an Apache-2.0 self-hosted AI gateway for OpenAI-compatible SDKs, MCP, A2A, CLI, and HTTP workflows — with the feature set you'd expect from a commercial LLM gateway: virtual keys with token budgets, exact + semantic response caching, reverse MCP governance (any OpenAPI 3 spec becomes governed MCP tools), and chat-native Prometheus/Langfuse observability.

GitHub stars CI Latest release Official MCP Registry: active Apache-2.0 license

Unified AI System turns a rough request into a structured coding prompt

Try It Without Installing

Open a ready-to-run prompt-enhancement example. It runs locally in the browser with no account, API key, or provider call.

Or run the full gateway in 60 seconds — no clone, no credentials:

docker run --rm ghcr.io/happy520ai/unified-ai-system/ai-gateway-service:0.5.0 pnpm gateway demo "Build a small API for my team" --enhance --profile coding

Why It Matters

  • Virtual keys (uai-) with daily/monthly token budgets, per-key rate limits, soft-budget alerts, spend attribution, and instant revocation — consumers never hold provider keys.
  • Exact + semantic response cache on the chat hot path with byte-identical SSE replay and per-tenant isolation.
  • Reverse MCP governance: aggregate upstream MCP servers (HTTP + stdio) behind one authenticated, audited, allow-listed surface; REST→MCP from any OpenAPI 3 spec.
  • The credential-free fake provider is the default; real-provider calls require explicit three-gate enablement and authorization.
  • A repeatable 16-attack live security regression (cross-tenant cache reads, tenant forgery, budget bypass...) must stay green for every release.

Use It With Codex

Install the repository plugin:

codex plugin marketplace add happy520ai/unified-ai-system --ref master

Or connect the pinned MCP container:

codex mcp add unified-ai-system -- docker run --rm -i ghcr.io/happy520ai/unified-ai-system/mcp-server:0.5.0

Project site | 中文 README | Release v0.5.0 | Good first issues

If the project saves setup time or gives an agent workflow a safer gateway, star the repository so more developers can discover and verify it.

Unified AI System is an engineering preview. Production readiness, L5 autonomy, and AGI are not claimed without independent evidence.

Pinned Loading

  1. unified-ai-system unified-ai-system Public

    Self-hosted AI gateway: OpenAI/Anthropic-compatible APIs, virtual keys with token budgets, exact+semantic response cache, reverse MCP governance (REST→MCP), Prometheus/Langfuse observability — fake…

    JavaScript 5 2