PentestCode is an autonomous, multi-agent penetration testing agent for the terminal, built as a hard fork of OpenCode and rebuilt for authorized offensive security.
Share
Follow PentestCode to get an email when its listing changes, such as new pricing or features. One digest a day at most, and you can unsubscribe anytime.
A lead coordinator plans the engagement and dispatches specialist subagents in parallel, and every agent reads and writes one persistent engagement state of hosts, vulnerabilities, credentials and attack paths.
- 01
Lead `pentest` coordinator with 13 specialist subagents, including recon, scanner, exploiter, AD identity, web app, critic and reporter
- 02
Shared engagement state of hosts, services, vulnerabilities, credentials, access and relationships that survives restarts
- 03
Attack-path ranking over the relationship graph
- 04
18 built-in pentest tools, with mandatory nmap, Nuclei, NetExec and Gobuster output parsers
- 05
Operator-gated scope with `/status`, `/vulns`, `/creds` and `/goal` commands
- 06
Human-readable `findings.md` log of every finding with timestamps
- 07
Works with 20+ LLM providers through the AI SDK, including Anthropic, OpenAI, Google, Bedrock and Ollama
- 08
Install with `npm install -g pentestcode-ai`, the install script or the AUR package
Open Source
PentestCode is free and open-source under the MIT licence. You install the single binary yourself and pay only for the LLM provider you connect; there is no hosted service, account or paid tier.
Best for
Penetration testers, red teams and CTF players who want an agent that runs and tracks a full authorized engagement from recon to report in their own terminal.
Not ideal for
Teams that need a supported commercial product, Windows-native installs or a web dashboard, and anyone without written authorization to test the target systems.
A capable, terminal-first take on agentic pentesting with a well-designed shared engagement state, still labelled beta by its author.
v0.2.7 was released on October 23, 2026, fixing a command-injection and an arbitrary-file-write issue and adding a screenshot tool, a subagent kill dialog and eight pentest methodology skills, after v0.2.6 added a reflector agent and cross-engagement knowledge store on September 2; the repository was trending on GitHub on November 16, 2026, with more than 1,000 stars.
As of
Is PentestCode free?+
Yes - PentestCode is Open Source. PentestCode is free and open-source under the MIT licence. You install the single binary yourself and pay only for the LLM provider you connect; there is no hosted service, account or paid tier.
Is PentestCode open source?+
Yes - PentestCode is open source. PentestCode is free and open-source under the MIT licence. You install the single binary yourself and pay only for the LLM provider you connect; there is no hosted service, account or paid tier.
Who is PentestCode best for?+
Penetration testers, red teams and CTF players who want an agent that runs and tracks a full authorized engagement from recon to report in their own terminal.
Who is PentestCode not ideal for?+
Teams that need a supported commercial product, Windows-native installs or a web dashboard, and anyone without written authorization to test the target systems.
What are the best PentestCode alternatives?+
The closest PentestCode alternatives on ai.dosa.dev are CodeRabbit, Qodo, Snyk Code - all listed under AI Code Review & Security.
Who makes PentestCode?+
PentestCode is developed by s0ld13rr. It is listed in the AI Code Review & Security category on ai.dosa.dev.