TencentOfficial·18 Agent Skills Included

AI-Infra-Guard

Scan AI infrastructure, agents, and skills for security risks

Scans AI services, MCP servers, agent skills, and LLMs for vulnerabilities, misconfigurations, and jailbreak weaknesses. Detects CVEs, prompt injection paths, data leakage, tool abuse, and supply chain risks without manual penetration testing. Generates evidence-based security reports with severity ratings and fix guidance so teams can harden AI systems faster.

6,103|568

↗

npx skills add Tencent/AI-Infra-Guard --all -g -y

Available:

Tells the AI agent how the platform is structured, how to build and run its scanners, and how to route scan tasks across its Go backend and Python security modules.

All Skills in This Repository (18)

Pure Emerald Level Indicators

📦 In Repo

Tencent

edgeone-clawscan

Audit OpenClaw environments for configuration, supply chain, and privacy risks.

owasp-asi

Classify AI agent security findings using the OWASP ASI 2026 framework.

tool-abuse-detection

Detect tool misuse and command injection by testing tool parameters through dialogue interactions.

data-leakage-detection

Detect sensitive information disclosure in AI models through escalating dialogue probes.

indirect-injection-detection

Detect indirect prompt injection attacks in external data sources.

authorization-bypass-detection

Detect authorization bypass vulnerabilities by simulating privilege escalation and cross-user access attempts.

memory-poisoning-detection

Detect persistent instruction injection in agent long-term memory through dialogue probes.

web-exfiltration-detection

Detect data exfiltration via URL path encoding and chained web navigation in AI agents.

cascading-failure-detection

Detect cascading failures and error propagation in multi-step agent workflows.

file-path-traversal-detection

Detects path traversal vulnerabilities in agent file upload, save, and extract flows.

agentic-supply-chain-detection

Detect supply-chain compromise risks in AI agents via dialogue-based probing.

human-agent-trust-exploit-detection

Detect social engineering and deceptive guidance in agent dialogue responses.

Official

Intermediate

Frequently Asked Questions

FAQPage Schema

How to install AI-Infra-Guard?▼

Run `npx skills add Tencent/AI-Infra-Guard --all -g -y` in your terminal to install all skills in this suite globally.

What does AI-Infra-Guard scan?▼

It scans AI infrastructure for CVEs and misconfigurations, audits MCP servers and agent skills for malicious code, tests agents for prompt injection and data leakage, and evaluates LLM jailbreak resistance.

How to check if an agent skill is safe before installing?▼

Use the included skill scanner to run a local static audit of any skill package; it flags credential theft, hidden network calls, and permission abuse in plain language.

Does AI-Infra-Guard work with Claude Code and OpenClaw?▼

Yes. The skills follow the universal SKILL.md standard and run in Claude Code, OpenClaw, Cursor, CodeBuddy, and similar agent clients.

Does my scan data leave my machine?▼

No. Static skill analysis runs fully locally, and optional cloud lookups send only minimal metadata like skill names and version numbers, which you can disable entirely.