vxcontrol/pentagi
Self-hosted multi-agent system for autonomous penetration testing
- Chart
- Not on a chart right now
- Stars
- 34.6k
What it is
PentAGI takes a target and goal in plain language, then coordinates specialized AI agents to plan and run 20+ security tools inside an isolated Docker sandbox. It stores past findings and generates vulnerability reports.
Who it is for
Security engineers, penetration testers, and red teams running authorized tests on their own infrastructure.
Why it is trending
It gained 140 stars today for a total of 34,494 stars, following its 2.2.0 release in October 2026. It stands out for combining self-hosted execution, 10+ cloud or local LLM providers, and persistent vector memory.
Main features
- Runs 20+ tools including nmap, Metasploit, and sqlmap in isolated Docker sandboxes
- Stores research results in PostgreSQL with pgvector and optional Neo4j knowledge graphs
- Supports 10+ cloud LLM providers alongside local models via Ollama or vLLM
- Provides a web UI plus REST and GraphQL APIs for automation and reporting
Summarized on November 18, 2026 from the README and repository data. It can be wrong; check the source before relying on it. Report a wrong summary