Tencent/AI-Infra-Guard — A full-stack AI Red Teaming platform securing AI ecosystems via OpenClaw Security Scan, Agent Scan,…

Description and use

**Summary**
Tencent/AI‑Infra‑Guard is a full‑stack red‑team platform that secures AI ecosystems by scanning OpenClaw, agents, skills, MCP, AI infrastructure and LLM jailbreaks. It also defines a standard Model Context Protocol (MCP) that lets AI assistants safely invoke real tools and data sources.

**Value**
The project bundles automated security checks with a unified protocol for connecting LLMs to external services, enabling teams to both harden their AI pipelines and expose trustworthy tool‑calling interfaces without reinventing the integration layer.

**Practical adoption path**

1. **Prototype** – Pull the Python SDK/CLI, run the built‑in scans on a sandbox model or agent to assess current risk.
2. **Integrate** – Deploy the MCP server (Docker or Helm chart) and register your internal tools via the provided schema; update your LLM’s prompt or tool‑calling wrapper to use the MCP endpoint.
3. **Scale** – Incorporate the scans into CI/CD pipelines (GitHub Actions, Jenkins, etc.) and monitor results through the built‑in dashboard or export to existing observability stacks.

**Production readiness**
With 3.6 k GitHub stars, recent commits (as of 2026‑05‑13), active forks, and a Python codebase that exposes clear API/SDK/CLI entry points, the project shows strong community traction and maintainability. The architecture is modular, the MCP server is container‑ready, and security scans are automated, making it suitable for a serious pilot in production environments—pending a final review of licensing and maintainer responsiveness.

Problem

Tencent/AI-Infra-Guard helps connect AI assistants to real tools and data through a standard protocol.

Outcome

High for an OSS candidate: recent activity, adoption, and ecosystem signals are strong enough for a serious pilot.

Workflow

• connect AI agents to tools
• ship Model Context Protocol servers
• standardize integrations

Evidence

A full-stack AI Red Teaming platform securing AI ecosystems via OpenClaw Security Scan, Agent Scan, Skills Scan, MCP scan, AI Infra scan and LLM jailbreak evaluation.

Source