# AI use cases for PentAGI (vxcontrol/pentagi)

Repo: [vxcontrol/pentagi](https://github.com/vxcontrol/pentagi) | Use cases: 3 | Page: https://dedupedai.com/use-cases/vxcontrol-pentagi/

PentAGI is an open-source, self-hosted multi-agent system for autonomous penetration testing, for authorized testing only. You describe a target and goal in plain language; specialized agents plan the work, run 20+ tools such as nmap, Metasploit and sqlmap inside an isolated Docker sandbox, keep what they learn in PostgreSQL with pgvector, and produce a report. It works with 10+ cloud LLM providers or local models through Ollama or vLLM, and has REST and GraphQL APIs.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Run an authorized pentest with a team of AI agents: https://dedupedai.com/use-cases/vxcontrol-pentagi/run-an-authorized-pentest-with-an-agent-team/
- Run PentAGI on local models so findings never leave your network: https://dedupedai.com/use-cases/vxcontrol-pentagi/run-pentagi-on-local-models/
- Start and collect pentest flows from your own tooling via API: https://dedupedai.com/use-cases/vxcontrol-pentagi/automate-pentagi-with-its-api/

## How to run an authorized pentest with a team of AI agents with PentAGI

Repo: [vxcontrol/pentagi](https://github.com/vxcontrol/pentagi) | Difficulty: Advanced | Time: 2 hours | Basis: From the README

Describe a target you are allowed to test, and let PentAGI plan, run tools in a sandbox and write the report.

**Who it is for:** A security engineer with a signed scope who wants the routine recon and testing done faster.

**The problem:** Manual pentests repeat the same recon and tool runs on every engagement.

**The outcome:** A flow report with findings and exploitation notes, viewable on the web or downloadable as Markdown or PDF.

### Steps

1. Install with the interactive installer (Docker Compose)
2. Configure an LLM provider
3. Log in to the web UI and start a flow with the target and goal in plain language
4. Watch agents run tools such as nmap and sqlmap in the Docker sandbox
5. Review and download the report

### Prompt to paste into your agent

```text
Help me install vxcontrol/pentagi with Docker Compose on an isolated lab server, configure <provider>, and run a flow against my own test VM at <lab address> only. Confirm the scope before starting.
```

### Caveat

Only test systems you own or have written permission to test; unauthorized scanning is illegal in most countries.

Risk, credentials: LLM keys and API tokens live in the deployment config; restrict access to the server.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/vxcontrol-pentagi/run-an-authorized-pentest-with-an-agent-team/

## How to run PentAGI on local models so findings never leave your network with PentAGI

Repo: [vxcontrol/pentagi](https://github.com/vxcontrol/pentagi) | Difficulty: Advanced | Time: 2 hours | Basis: From the README

Point PentAGI at Ollama or vLLM so prompts and results stay on your hardware.

**Who it is for:** A consultancy whose clients forbid sending engagement data to cloud AI providers.

**The problem:** Pentest data is sensitive and cloud LLMs may be off limits.

**The outcome:** PentAGI running against a local model with execution monitoring and task planning to help smaller models.

### Steps

1. Set up Ollama or vLLM on a GPU server
2. Configure PentAGI's custom or Ollama provider
3. Enable execution monitoring and task planning for smaller open models
4. Run a lab flow and compare results with a cloud model

### Prompt to paste into your agent

```text
Configure vxcontrol/pentagi to use my local <Ollama or vLLM> endpoint with <model>, enable execution monitoring, and run a short flow against my lab VM. Report which steps the local model struggled with.
```

### Caveat

Authorized targets only; smaller local models may miss findings a larger model would catch.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/vxcontrol-pentagi/run-pentagi-on-local-models/

## How to start and collect pentest flows from your own tooling via API with PentAGI

Repo: [vxcontrol/pentagi](https://github.com/vxcontrol/pentagi) | Difficulty: Advanced | Time: 1 hour | Basis: From the README

Use the REST or GraphQL API with a Bearer token to trigger flows and pull results into your pipeline.

**Who it is for:** A security team that wants scheduled lab scans feeding its tracker.

**The problem:** Clicking through a UI does not scale to recurring checks.

**The outcome:** Scripts that start PentAGI flows and fetch reports automatically.

### Steps

1. Create an API token in PentAGI
2. Call the REST or GraphQL API with Bearer token authentication
3. Start a flow for an authorized target and poll for completion
4. Fetch the report and file findings in your tracker

### Prompt to paste into your agent

```text
Write a script that uses vxcontrol/pentagi's REST API with a token from an environment variable to start a flow on <authorized lab target>, wait for it, and save the Markdown report.
```

### Caveat

Authorized targets only.

Risk, credentials: API tokens can start attacks on any configured target; store them as secrets and rotate them.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/vxcontrol-pentagi/automate-pentagi-with-its-api/