DarkMoon CLI
DarkMoon runs an autonomous AI pentest of 50 agents on your infrastructure, with the command and output for every finding
- 1,217 ★ on the repository
- Last commit
- GPL-3.0
Broad access
Why this level
- Agents exploit vulnerabilities on the target: running it against someone else's system without permission breaks the law and can lead to liability
- Exploitation can crash a service or change data, so test a staging environment and leave a live system with no rollback alone
- Commands run in a container with pentest tools, and the provider key is stored on the machine in `.opencode.env`
Install
- Works in
Manual install
git clone https://github.com/ASCIT31/Dark-Moon.git && cd Dark-Moon && ./install.shDocker and Docker Compose are required. The script configures the model provider interactively.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Agents exploit vulnerabilities on the target: running it against someone else's system without permission breaks the law and can lead to liability
- Exploitation can crash a service or change data, so test a staging environment and leave a live system with no rollback alone
- Commands run in a container with pentest tools, and the provider key is stored on the machine in `.opencode.env`
This is third-party code. Review the repository files before installing.
What it does
DarkMoon is a command line tool for autonomous penetration testing. You name an authorized target with `./darkmoon.sh "TARGET: example.com"`, and the agents run reconnaissance, plan the attack, exploit vulnerabilities and write down the result. Every finding carries the exact command and raw output, so each finding can be rechecked by hand. The kit has 50 specialist agents for web applications, APIs, cloud, Active Directory, Kubernetes, CI/CD and AI systems per the OWASP LLM Top 10. Each agent's methodology is a plain Markdown file you can read and edit. Commands run in a Docker container through an MCP server that acts as the boundary between the model and the real tools. Installation is the `./install.sh` script: it asks for a model provider, cloud or local, and builds the whole stack. Anthropic, OpenAI, OpenRouter, Ollama and llama.cpp work. A privacy gateway swaps real addresses, hosts and credentials for placeholders while the model reasons. Integrations exist for GitHub Actions, GitLab and Jenkins, plus plugins for VS Code and JetBrains.
Who it is for. For security specialists and development teams who test their own infrastructure or systems with the owner's written permission.
Good fit when
- You need a first automated check of your own environment: web, API and Kubernetes with a report per finding
- Data must stay inside the perimeter: a local model through Ollama and a privacy gateway instead of a cloud provider
- A pentest must run in CI: GitHub Actions, GitLab and Jenkins integrations, with the build failing at a chosen severity
- The team wants to see the agents' methodology and adapt it, since it lives in Markdown
Not a fit when
- The target is not yours and you have no owner permission: scanning and exploiting someone else's system is illegal in many countries
- You need a web interface, a campaign scheduler or automatic fix pull requests: these are paid Pro features
- You need to test production during business hours with no rollback: exploitation can disrupt the service
- You need an exact reproducible result without review: the agent asserts exploitation, and conclusions need manual checking
Example request
Run DarkMoon on our staging.example.com test environment and show the vulnerabilities found with commands and outputLimitations
Running it needs Docker with Docker Compose and a model provider key or a local model. The result depends on the model: a weak local model finds fewer vulnerabilities, and the provider choice decides where pentest data goes. With a cloud provider command output leaves your network, and only the privacy gateway swap hides the real values. The agent itself confirms exploitation, and machine-verified fix retesting exists only in the paid edition. The GPL-3.0 license applies to the open part. The last commit dates from October 2026 and the project is under active development.
How to disable. Stop the stack with `docker compose down` in the project directory and delete the repository folder. Provider keys live in `.opencode.env`, so remove that file.
FAQ
How do I install DarkMoon?
Clone the repository, enter it and run `./install.sh`. The script asks for a model provider and builds the stack in Docker. The `--init` flag forces reconfiguration.
How do I run a pentest with DarkMoon?
Run `./darkmoon.sh "TARGET: example.com"`. Follow progress with `./darkmoon.sh --log <session_id>`, and the report gives the exact command and raw output for every finding.
Can DarkMoon run on a local model?
Yes. Ollama and llama.cpp are supported, so the model runs on your machine and pentest data does not go to a cloud provider.
What is in the free edition and what is in Pro?
The open edition has the command line engine and the agents. The web dashboard, recurring campaign scheduler, report export and automatic fix pull requests belong to the paid Pro edition.
Is it legal to run DarkMoon on any website?
No. The tool is meant for targets you may test: your own environment or a system with the owner's permission. Running it on someone else's infrastructure without consent breaks the law in many jurisdictions.
Discussion
No questions or reviews yet.
Related
CLI toolsSelf-hosted AI pentesting framework: attack surface graph, Kali sandbox, human approval gates and an MCP server you can drive from Claude Code
An authorized bug-bounty suite: 50 subagents, 26 commands and 2 bug-bounty-platform MCP servers for Claude Code, Codex and five more agents
A CLI agent for authorized pentesting: from a natural-language brief it runs recon, finds and verifies vulnerabilities and prepares a report
HackAgent tests your AI model or agent for jailbreaks, prompt injection and tool misuse and reports a measured attack success rate
A pack of about fifty Claude Code subagents for authorized pentesting: planning, recon, exploit research and chaining, post-exploitation, detection and reportin
Open-source deception framework: decoy SSH, HTTP, database and MCP services in YAML, optional LLM responses, attacker activity logged as evidence
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| DarkMoon (this card) | no | 1,217 | ||
| Strix | no | 75,261 | another autonomous pentest agent with its own toolset and report format | |
| HexStrike AI MCP Agents | no | not required | 13,546 | an MCP server of security tools for your own agent, not a ready engine with 50 agents |
README in short
DarkMoon is an open autonomous pentest engine with a Docker-based command line launch, 50 specialist agents and an MCP layer that limits what the model can do. CI and IDE integrations exist, and extended features are in the paid edition.