Gitspawn Scan — AI skill for Claude Code
Scan a folder before your AI coding agent opens it.
How to install Gitspawn Scan
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open shaojun-666/gitspawn-scan and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Gitspawn Scan does
Scan a folder before your AI coding agent opens it. Finds the git config, hooks and agent settings that execute code the moment an agent touches the directory.
Alternatives in AI
- Swift Agent Skills — A curated directory of open-source AI agent skills for Swift and Apple platform development 1.5k ★
- Coffee CLI — A clean AI terminal workspace 180 ★
- Engram — The context spine that 10x's every AI coding session 141 ★
README
gitspawn-scan
[English](README.md) | [简体中文](README.zh-CN.md)
[](https://github.com/shaojun-666/gitspawn-scan/actions/workflows/test.yml)
**A folder can run code at you before your AI agent finishes opening it. Check first.**
npx github:shaojun-666/gitspawn-scan ~/Downloads/that-repo-someone-sent-me
One command. No dependencies, nothing to install, no npm package in the middle.
What happens without it
Someone sends you a zip. You unpack it, open it up, and tell Claude Code to take a look at the project.
At that moment a program may already have run on your machine, and you did not write it.
You did not click anything, and nothing asked you for permission. What triggered it was a `git status` — the first thing an agent runs to orient itself. That git command reads the repository's own `.git/config` and executes a program named in there.
The awkward part is where the execution happens: inside git, below the agent's tool layer. So there is no tool call for the agent to approve, no sandbox in the way, and in some agents it fires before the workspace-trust prompt is even answered. By the time you think to check, there is nothing left to check.
Where this came from
Manifold Security disclosed the class in September 2026 and named it GitSpawn ([write-up](https://thehackernews.com/2026/09/malicious-git-configs-can-make-claude.html)). Eight findings across seven agents: Claude Code, Codex, Cursor, Goose, Hermes Agent, Qwen Code and Grok Build. Several were still unpatched weeks later.
The mitigation they published was to look by hand:
git config --get core.fsmonitor
Sound advice. It just covers one key, and one repository at a time.
gitspawn-scan does the whole check: every configuration key in that family, plus a category nothing else looks at — the agent's own configuration files.
Clone is not the risky path
Worth knowing, becau
Related Skills
Agent Settings Backup Script
Git-versioned backup tool for AI coding agent config folders (Claude Code, Cursor, Codex, etc.) with size-base
Fireraven Agent Hooks
FireGuard hooks for AI agents: protect Cursor, Claude Code, Windsurf, Devin, and Copilot Studio from secret le
AI Git Hooks
AI-powered git hooks - auto-review diffs, generate commit messages, scan for secrets. Supports Claude, OpenAI,
Council Config
View and manage LLM Council configuration settings.
Crew Config
Configure CodeCrew settings - model tiers, default mode, token budgets, and more. Subcommands: show, set, rese
Claude Auto API
Claude Code settings.json auto-config tool to quickly switch API_KEY, AUTH_TOKEN, and model configs across mul
Related Agents
Component Reviewer
Expert component reviewer for Claude Code Templates. Use PROACTIVELY when adding or modifying components in cl
Light Security ClaudeSettingsAudit
Audit Claude Code config files (settings.json, settings.local.json, .mcp.json) for unsafe or malicious content
Claude Code Currency
Claude Code setup currency expert. Researches the latest official Claude Code config practices (subagents, set