awesome-agent-security
Curated list of AI agent security tools, papers and datasets.
What it is
Curated list of AI agent security tools, papers and datasets: runtime controls, sandboxes, scanners for MCP servers and skills, static analysis, red teaming, incident records, threat models and standards.
An AI agent is a model that can read untrusted content and call tools with real privileges. Securing one means controlling what it may touch, isolating what it runs, scanning what it loads, and measuring how it fails. Every entry below exists on GitHub and is maintained; descriptions are the projects' own words, shortened. Maintained by Muhammad Basit Ali; projects by the maintainer are listed in their sections like any other entry. See contributing.md for the inclusion criteria.
Links
Topics: agent-security, agentic-ai, ai-agents, ai-security, awesome, awesome-list, datasets, hacktoberfest, llm-security, mcp, mitre-atlas, model-context-protocol, owasp, prompt-injection, red-teaming, sandboxing, security-tools, threat-modeling