CompBox CLI - Install, update, and manage Skill Packs
Soft block: AI-agent control surface
Warn by default; block when configured. Any existing user instructions in that file are lost, allowing the package to replace the agent's instructions.
AI security review
scanned · by lpm-firewall-ai
The cloud workspace runner unconditionally replaces an agent instruction file during job setup. The package's intended agent setup does not make overwriting existing user instructions safe.
The `compbox workspace run` path processes a cloud workspace job.
Impact
Any existing user instructions in that file are lost, allowing the package to replace the agent's instructions.
Mechanism
It writes generated CompBox routing instructions directly to `$HOME/.claude/CLAUDE.md` without preserving existing contents.
Policy narrative
When a cloud workspace job runs, setup writes CompBox routing instructions directly over `$HOME/.claude/CLAUDE.md`. Because the write replaces the file rather than preserving existing instructions, user-authored Claude instructions can be erased without a specific request.
Evidence
index.js$HOME/.claude/CLAUDE.md
Decision evidence
public snapshot
AI identified this as a dangerous AI-agent capability at 88.0% confidence with low false-positive risk.
Evidence for policy risk
`compbox workspace run` is the cloud workspace job runner.
During cloud job setup, it writes generated instructions directly to `$HOME/.claude/CLAUDE.md`, replacing any existing contents.
The generated instructions route prompts to CompBox agents; the write does not preserve or append to the existing instruction file.
Evidence against
The package is a CompBox CLI, and this behavior is part of its cloud workspace agent setup.
The generated instructions describe CompBox agent routing and prohibit pushing to the default branch.
Avoid installing @mycompbox/compbox-cli@1.21.19. Remove it from direct dependencies and check your lockfile for transitive copies.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.
If this version ran, check your AI agents' configuration and instruction files (the files this report lists) for changes you did not make, and remove them. This finding does not by itself establish malware intent.