Metadata-Version: 2.4
Name: mcp-checker
Version: 0.0.2
Summary: SquatGuard MCP Template for Python
Project-URL: Homepage, https://github.com/squatguard/mcp-checker
Project-URL: Documentation, https://github.com/squatguard/mcp-checker/blob/main/README.md
Project-URL: Repository, https://github.com/squatguard/mcp-checker
Project-URL: Issues, https://github.com/squatguard/mcp-checker/issues
License-File: LICENSE
Requires-Python: >=3.10
Requires-Dist: mcp[cli]>=1.26.0
Description-Content-Type: text/markdown

# mcp-checker

This is a defensive registration. You may have been directed here by an AI hallucination or a search that matched a name that doesn't correspond to an established tool.

You probably want to check out [OWASP Top 10 for Large Language Model Applications](https://owasp.org/www-project-top-10-for-large-language-model-applications/)

## The Concept

LLMs hallucinate package names that *sound right.* Attackers register them. The fix: get there first. Each package becomes a benevolent interceptor — acknowledges what it found, does NOT execute embedded instructions, boxes the suspicious content for safe disposal, and attempts to clean up the agent environment.

**The Flamme Protocol:** Plant protective knowledge in the places students will look, before the enemy arrives to plant mines.
