Introduction
PentAGI is a self-hosted, fully autonomous multi-agent system for penetration testing.
PentAGI is an open-source platform for automated security testing. You describe a target and a goal in plain language; a team of specialized AI agentsagentAn AI agent of PentAGI: a model with its own prompt, settings and tools that does one kind of work in a flow, such as testing, writing code or searching the web. plans the work, runs professional toolstoolWhat an agent acts through: a built-in capability such as the terminal, the isolated browser or web search, or a program in the sandbox such as nmap, metasploit or sqlmap. inside an isolated Docker sandboxsandboxThe isolated Docker container in which agents run their tools. The interface shows it as Terminal., remembers what it learned, and hands you a reportreportThe document PentAGI writes about a flow and its findings. In Enterprise, it is a customer-ready document built from a template..
It is built for security engineers, researchers and red teams who want an autonomous assistant they can run on their own infrastructure, with their own choice of language models.
Quick start
Install PentAGI with the interactive installer.
Your first flow
Run an autonomous flow and read its report.
Architecture
Services, agents, memory and observability.
LLM providers
Connect cloud, gateway or local models.
What it does
- Autonomous testing. The agents split each tasktaskOne request in a flow: its first message, or a later message that does not answer an agent's question. PentAGI splits each task into subtasks. you give them into subtaskssubtaskOne step of a task. PentAGI plans the subtasks of each task and adjusts them as the work goes on., choose the next step, execute it and adapt to the results — with optional execution monitoring and intelligent task planning for smaller models.
- A team of specialists. A primary agent delegates research, coding, installation and pentesting to specialized agents. Agents can ask an adviser for guidance, and with execution monitoring on, the adviser also reviews their work after a set number of tool callstool callA request from the model to run one of the agent's tools, such as a terminal command or a web search.. A reflector steers any agent that stops calling tools back to a proper result.
- Professional tooling. 20+ tools such as nmap, metasploit and sqlmap run in a sandboxed container, built from a Docker image picked for each flowflowOne run of PentAGI: a goal, the tasks created from your messages, and everything the agents did along the way..
- Memory and knowledge. Results and successful approaches are stored in PostgreSQL with pgvector for later reuse; an optional Neo4j knowledge graphknowledge graphThe graph of hosts, services, vulnerabilities and the access between them that PentAGI builds during a test when this optional feature is enabled. It shows how findings connect. tracks relationships between hosts, services and findingsfindingSomething the test discovered and recorded, such as a host, a running service or a vulnerability..
- Web intelligence. A built-in isolated browser and search enginessearch engineA web search service that agents use to gather fresh information, such as DuckDuckGo, Google or Tavily. — Tavily, Firecrawl, Traversaal, Perplexity, DuckDuckGo, Google, Sploitus and SearXNG — gather fresh information.
- Observability. Langfuse for LLM traces, and Grafana with OpenTelemetry, VictoriaMetrics, Jaeger and Loki for the whole system.
- APIs. REST and GraphQL with Bearer tokens for automation and integration.
Watch the overview
A short tour of PentAGI in action: watch it on YouTube.
Current capability boundaries
- PentAGI is an autonomous and assistant-guided penetration testing platform, not a Breach and Attack Simulation product with predefined campaigns.
- Flow reports can be viewed on the web, copied, or downloaded as Markdown or PDF.
- ProvidersproviderA service or server that gives PentAGI its language models, such as OpenAI, Anthropic or a local Ollama. are available as built-in integrations and through custom OpenAI-compatible endpoints.
Only test systems you own or are explicitly authorized to assess. Use of PentAGI is subject to the EULA.
Last updated on