Skillget

claude-code-project-boundary

skill
Security Audit
Fail
Health Pass
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 36 GitHub stars
Code Fail
  • rm -rf — Recursive force deletion command in hooks/guard.sh
  • rm -rf — Recursive force deletion command in tests/test_guard.sh
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This is a security plugin for Claude Code that creates a protective boundary around your project directory. It permits potentially destructive file operations (like `rm` or `chmod`) only within the project scope, while automatically blocking them outside the directory to prevent accidental system-wide damage when running in a highly permissive mode.

Security Assessment
The tool is designed entirely around executing and filtering shell commands, which is its core purpose. It does not make external network requests, nor does it contain hardcoded secrets. However, the automated scan flagged the presence of `rm -rf` commands within the `hooks/guard.sh` and `tests/test_guard.sh` files. This is expected given the tool's objective of regulating file deletion, and these commands are restricted to the project boundary. While the tool drastically reduces the risk of AI-driven system destruction, relying on any shell-parsing safety net—especially one guarding a `dangerouslySkipPermissions` environment—inherently carries a slight risk of bypass. Overall risk is rated as Medium.

Quality Assessment
The project demonstrates strong health and maintenance indicators. It is licensed under the standard MIT license, meaning it is open and safe to integrate. The repository is highly active, with the most recent push occurring just today. Additionally, it has garnered 36 GitHub stars, showing a reasonable degree of community trust and peer validation for a highly specific developer tool.

Verdict
Use with caution — the tool is well-maintained and conceptually sound, but it acts as a regex-based safety net for inherently risky execution modes.
SUMMARY

Claude Code plugin: scope-aware protection — allows destructive ops (rm, mv, chmod) within project, blocks them outside. Built for dangerouslySkipPermissions mode.

README.md

Project Boundary — Claude Code Plugin

Allows destructive operations within your project but blocks them outside the project directory. Built for dangerouslySkipPermissions mode where Claude doesn't ask — this plugin is your safety net.

How it differs from existing plugins

  • claude-code-safety-net — blocks rm everywhere; Project Boundary allows it inside the project so refactoring works normally.
  • destructive-command-guard — only distinguishes /tmp vs everything else; Project Boundary uses $CLAUDE_PROJECT_DIR as the actual boundary.
  • claude-code-damage-control — requires manually listing protected paths; Project Boundary automatically protects everything outside the project.

What it does

Always blocked (regardless of location)

Command Example
git push --force / -f git push --force origin main
git reset --hard git reset --hard HEAD~1
git checkout . git checkout .
git clean -f git clean -fd
DROP TABLE / TRUNCATE TABLE DROP TABLE users;
rails db:drop / db:reset rails db:drop
rake db:drop / db:reset rake db:drop
mkfs.* mkfs.ext4 /dev/sda1
dd if= dd if=/dev/zero of=/dev/sda

Boundary-checked (allowed inside project, blocked outside)

Operation Inside project Outside project
rm, rm -rf Allowed Blocked
mv (source and destination) Allowed Blocked
cp (source and destination) Allowed Blocked
ln (source and target) Allowed Blocked
chmod / chown Allowed Blocked
> / >> redirect Allowed Blocked
tee / tee -a Allowed Blocked
curl -o / curl --output Allowed Blocked
wget -O / wget --output-document Allowed Blocked
find -delete / find -exec rm Allowed Blocked

Always blocked (unsafe to inspect)

Command Reason
bash -c "..." / sh -c "..." Nested shell — cannot inspect inner command
eval '...' Cannot safely parse evaluated code
Piping to sh / bash Inner commands invisible to guard
xargs rm/mv/cp/... Arguments cannot be validated

Additional protections

  • Chained commands — splits on ;, &&, ||, | and checks each sub-command independently
  • cd tracking — cd /tmp && rm -rf something is blocked because cd left the project
  • sudo prefix — stripped before checking, so sudo rm /etc/passwd is still blocked
  • find options — handles -L, -H, -P before the search path
  • Path traversal — .. segments are resolved before boundary check
  • ~ and $HOME expansion — rm ~/file and rm $HOME/file are correctly detected as outside-project
  • Symlink resolution — handles macOS /var → /private/var and similar

Known limitations

  • Paths with spaces in project directory name are not fully supported (space-based argument splitting)
  • $() subshells and backtick substitution inside arguments are not expanded

Install

Direct:

claude --plugin-dir /path/to/claude-code-project-boundary

From marketplace:

/plugin marketplace add davepoon/buildwithclaude
/plugin install project-boundary@buildwithclaude

How it works

A pure-bash PreToolUse hook. Splits chained commands, resolves target paths (handling symlinks, .., ~, $HOME), and compares against $CLAUDE_PROJECT_DIR. Dependencies: bash + jq.

Testing

bash tests/test_guard.sh

125 tests covering all guard scenarios. CI runs on Ubuntu and macOS.

License

MIT

Reviews (0)

No results found