Practical overview
When it fits
Use Pentestgpt when you need a self-hosted command-line framework for an authorized assessment or CTF workflow and can provide the required model backend, CLI authentication or provider API credentials.
Who it is for
- Security teams conducting authorized penetration testing
- Researchers studying AI-assisted security workflows
- CTF participants and learners
- Developers building self-hosted LLM-assisted assessment workflows
Documented uses
- Organize reconnaissance, exploitation and walkthrough stages for CTF work
- Move a penetration test from asset discovery through vulnerability identification to reporting
- Preserve and resume testing sessions
- Guide an interactive security-testing session through a Pentesting Task Tree
- Use Claude Code, Codex or configured LLM providers within a CLI workflow
Requirements & limits
The repository is public and MIT-licensed. Local setup requires Python 3.12+, uv and authenticated Claude Code or Codex CLI access for the autonomous workflow; legacy mode uses configured provider API keys or local Ollama.
- The autonomous pipeline is documented for Claude Code and Codex, while the interactive legacy mode supports a wider provider set.
- The framework requires a local CLI or Docker environment rather than a hosted web interface.
- The penetration-test workflow and CTF workflow use different documented stage sequences.
- Session persistence is available for continuing testing work.
- The repository identifies Python as its primary language and uses an MIT license.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install Pentestgpt for local use
Set up the repository and install its Python dependencies.
Before you start
- Python 3.12 or newer
- uv Python package manager
- A cloned copy of the PentestGPT repository
Steps
Clone the PentestGPT repository.
Change into the repository directory.
Run the command below, which runs the command below.
make installuv sync
Expected outcome
The documented local installation process has been completed.
Guide limitations
- This setup recipe covers installation only; it does not establish that a target run will succeed.
- Running the autonomous workflow additionally requires authenticated Claude Code or Codex CLI access.
Run a documented penetration-test workflow
Start Pentestgpt in penetration-test mode against an authorized target.
Before you start
- An installed Pentestgpt environment
- An authorized target
- Authenticated Claude Code or Codex CLI access
- A target address
Steps
Provide the target with the
--targetoption.Set
--mode pentest.Select the documented backend and model when using the Docker flow.
Allow the pipeline to proceed through asset discovery, vulnerability identification and reporting.
Expected outcome
The penetration-test pipeline is started with the documented staged sequence.
Guide limitations
- The framework's documented pipeline is for authorized testing only.
- The recipe describes the supplied workflow stages, not the results of a particular assessment.
- The autonomous pipeline supports Claude and Codex backends; broader provider selection is documented for legacy mode.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 2 October 2026
- Original website title
- GreyDGL/PentestGPT · Official homepage title · retained in its source language