Penthera plugin
Offline security scanner for code and dependencies with a plugin and MCP: 132 rules, findings with evidence at file:line.
- 106 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The scanner reads the whole project and, once confirmed, calls the running app, and fixes are written to files after approval.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/pentheraDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The scanner reads the whole project and, once confirmed, calls the running app, and fixes are written to files after approval.
This is third-party code. Review the repository files before installing.
What it does
Penthera checks an app, its dependencies and the settings of AI tools right on the developer's machine. 132 readable rules make the decisions and every finding carries evidence with a file and line. In `--ai` mode a model only proposes hypotheses and the rules verify each one: unverified ones are listed apart and never fail CI. In a terminal `npx penthera` detects the stack, asks what to check and returns a grade per area with the three top fixes. A running app is scanned only after you confirm it is yours. Fixes go through `penthera fix --plan` and are approved by hand. In Claude Code there is a plugin with a skill that runs the CLI, or an MCP server with six read-only tools. `penthera ui` opens a dashboard with a plain-language verdict and a triage matrix.
Who it is for. Developers and owners of small products, including ones built in Cursor, Lovable or Bolt, who need a security check before publishing.
Good fit when
- Before shipping an app you want to find leaked secrets and risky dependencies.
- An agent built a project in one evening and its code needs an independent check.
- You want a security check in CI with a clear exit code.
- A suspicious npm package should be checked before installing.
Not a fit when
- You need a human penetration test of a live system: Penthera finds known classes of problems.
- The target is someone else's site without the owner's permission.
- You want model conclusions that no rule has verified.
Example request
Run Penthera on this repository, show the three main problems and propose a fix plan without applying it.Limitations
Penthera finds known classes of problems and does not replace a security specialist. It needs Node.js 22 or newer. A running app is scanned only after you confirm it belongs to you. The `--ai` mode sends data to the chosen model, and findings not confirmed by rules never change the exit code. Results of public tests on leaked secrets and malicious packages are in the README: each dataset has its own profile, so figures from one cannot be carried to another. Fixes are applied only after your approval, and nothing is committed or pushed.
How to disable. Remove the plugin with `claude plugin uninstall penthera@penthera` or the MCP entry with `claude mcp remove penthera`. Running through `npx` leaves nothing in the project except settings you saved yourself.
FAQ
How do I install Penthera in Claude Code?
Run `claude plugin marketplace add danoszz/penthera`, then `claude plugin install penthera@penthera`. For the MCP server alone use `claude mcp add penthera -- npx -y penthera@2.4.0 mcp`.
Is installation needed for a first run?
No, `npx penthera` in the project folder and Node.js 22 are enough. No account is required.
What does the model do with `--ai` on?
It proposes problems no rule describes and the rules check each hypothesis. Unverified findings are listed apart and never fail CI.
Which tools does the MCP server expose?
Six read-only tools. The agent makes fixes itself through the skill or `penthera fix --plan`.
Can the scan run in a container?
Yes: `docker build -t penthera .`, then `docker run --rm penthera https://your-app.example`.
Discussion
No questions or reviews yet.
Related
PluginsA security scanner for MCP servers and agent skills that looks for prompt injection, secret leaks and swaps after approval.
A code security plugin: finds vulnerabilities, challenges its findings, proves exploitability and fixes with a verified patch
A local scanner flagging malicious packages and risky permissions in dependencies, agent instructions and MCP configs.
A SAST scanner with cross-file dataflow analysis and an MCP server: the agent sees call graphs, taint sources and sinks
Security scanner for MCP servers that looks for hardcoded keys, tool poisoning and injection flaws in repository code.
Prowler plugin for Claude Code
Prowler for Claude Code
A plugin that takes a cloud account toward CIS, PCI-DSS and other frameworks: reads Prowler Cloud reports, fixes failures and proposes a rescan
README in short
The developer works in the open and publishes results on third-party datasets, including poor ones. The README has a setup table for about ten agents and a comparison with Snyk, Socket, Semgrep, ZAP and GitHub.