pentagi
Fully autonomous AI Agents system capable of performing complex penetration testing tasks
PentAGI is a fully autonomous AI agent system for penetration testing that leverages multi-agent architecture, LLM-powered reasoning, and 20+ professional security tools. It operates in isolated Docker containers, delegating reconnaissance, exploitation, and reporting tasks to specialized AI sub-agents (researcher, developer, executor). The system features long-term memory with pgvector, optional knowledge graph integration via Graphiti/Neo4j, and comprehensive monitoring through Langfuse and OpenTelemetry. Supports 10+ LLM providers including OpenAI, Anthropic, Gemini, AWS Bedrock, DeepSeek, and local Ollama/vLLM deployments.
Key Features
Use Cases
- 01Autonomous web application vulnerability assessments with multi-stage attack chains
- 02Penetration testing in air-gapped environments using local LLM inference (vLLM/Ollama)
- 03Security research workflows requiring coordinated tool execution and exploit development
- 04Continuous security monitoring with persistent memory of successful attack patterns
- 05Complex threat modeling with knowledge graph-backed relationship tracking
- 06Batch penetration testing via REST/GraphQL APIs integrated into CI/CD pipelines
Related Agents
View morehermes-agent
The agent that grows with you
agency-agents
A complete AI agency at your fingertips - From frontend wizards to Reddit community ninjas, from whimsy injectors to reality checkers. Each agent is a specialized expert with personality, processes, and proven deliverables.
openinterpreter
A coding agent for open models like Kimi K3
cline
Autonomous coding agent as an SDK, IDE extension, or CLI assistant.
pentagi — FAQ
What is PentAGI?+
PentAGI is an autonomous AI agent system for penetration testing that uses multiple specialized AI agents (researcher, coder, executor) to automatically discover, exploit, and report security vulnerabilities. It runs all operations in isolated Docker containers with 20+ professional security tools.
How do I install PentAGI?+
Download the interactive installer for your platform (Linux/Windows/macOS), run it with Docker access (sudo or docker group membership), and follow the TUI prompts to configure LLM providers, search engines, and security settings. Alternatively, use docker-compose with a manually configured .env file.
Which LLM providers does PentAGI support?+
PentAGI supports OpenAI, Anthropic, Google Gemini, AWS Bedrock, Ollama (local), DeepSeek, GLM, Kimi, Qwen, MiniMax, and custom OpenAI-compatible endpoints (vLLM, LiteLLM). Per-agent model assignment is configurable via YAML provider configs.
What API keys or prerequisites are required?+
At minimum, you need one LLM provider API key (e.g., OpenAI, Anthropic, or a local Ollama server). Optional: search engine keys (Google, Tavily, Firecrawl, Perplexity), Langfuse for LLM observability, Neo4j for knowledge graph. Docker and 4GB RAM are required.
Is PentAGI free to use?+
The MIT-licensed code is free. LLM provider costs depend on your choice: paid cloud APIs (OpenAI, Anthropic) or free local inference (Ollama, vLLM). VXControl Cloud Services (threat intelligence, premium features) require a separate license key.
Which AI clients work with PentAGI?+
PentAGI is a standalone autonomous agent system accessed via web UI (React frontend) or REST/GraphQL APIs. It is not an MCP server or plugin for Claude Desktop/Cursor; instead, it orchestrates its own specialized sub-agents internally.
How do I install pentagi?+
Open the source repository on GitHub and follow its README. pentagi is a agent — MCP Agents Market links you directly to the official repo.
Is pentagi free?+
pentagi is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.