Plugin · 9 included skills
Master Security Toolkit
Before trusting a package, plugin, MCP server or AI prompt design, check what it can access and where it came from. Master Security Toolkit helps you review evidence, test risks and verify fixes.
Evidence first. No safety certificates.
What can you do with it?
- 01
Decide whether to install something
Review a package, repo, plugin or MCP server before you run it, with a clear verdict, confidence and safer path.
- 02
Find AI instruction risks
Assess prompts, retrieval and tools for direct and indirect prompt injection, and what an attacker could actually reach.
- 03
Fix it and prove the fix
Turn a finding into layered mitigations and repeatable regression tests with honest residual risk.
Start with one artifact.
Share the local artifact or exact source revision, how you plan to use it and the decision you need. Ask what the toolkit can do if you do not know which review fits.
Before I install [package, plugin or MCP], review it for security risk. Read its manifest and scripts, check its permissions and where it came from, and give me a verdict with your confidence, what you inspected and anything you couldn't check.
Bring: The artifact or exact source revision and your intended use. Expect: A verdict, confidence, evidence and a safer path.
Explore 8 example prompts →Add it to your AI app.
Download the plugin ZIP, then follow the setup for your host. Start a fresh session and try one task.
SHA-256 checksum file for this plugin ZIP.
Uploading a ZIP into an ordinary web chat does not install the plugin.
The reviews work from the artifacts you supply. The local scanner needs Python 3; publication preflight needs Python 3.10 or newer; Node.js is needed only for one coverage-audit validator. Check the requirements.
Technical details, checksums and historyChoose a skill for your task.
Start with Software Security Review for an install decision, or name the specialist for the surface you're reviewing.
Before you install
Software Security Review
Review an app, package, script, installer or repo before you install or run it.
See this skillSoftware Provenance
Verify that a download or release is authentic and unmodified.
See this skillAgent Artifact Scanning
Scan a local skill or plugin package for risky instructions, permissions and secrets.
See this skillMCP Security Review
Review an MCP server or plugin's launch command, tools, permissions and auth before trusting it.
See this skillPermission and Data Review
Check what an install or agent action can touch: permissions, persistence and private data.
See this skill
AI instruction security
Instruction Security Review
Assess prompt-injection risk across prompts, retrieval, tools and memory.
See this skillIndirect Injection Review
Review content an AI reads, like web pages, documents and tool output, for hidden instructions.
See this skillSecurity Regression Tests
Turn a confirmed finding into layered mitigations and repeatable pass/fail tests.
See this skill
Applications you own
Automatic guidance: hooks
Version 0.3.0 registers no automatic hooks. You can use the included skills by asking your assistant for the task you want to complete.
Included scanner and optional catalog
No hook, background service or automatic gate starts with this toolkit. It includes a pinned local scanner and an optional catalog launcher.
Prompting for newcomers
Build basic prompting skills before exploring specialized tools. This beginner course teaches prompting and draft revision through a social-media project.
Learning Studio course: ChatGPT for Beginners: Learning to Prompt →Setup and compatibility
- Your AI host
- Use a host with file tools that can read the artifact you're reviewing. Read setup details
- Local runtimes
- Python 3 for the scanner and catalog launcher, Python 3.10 or newer for publication preflight, and Node.js only for the coverage-audit validator. Read setup details
- Separate outcomes
- Source review, controlled testing, live tool access and installation are separate results. Target access and test execution need their own evidence. Read setup details