AI Security Review
scanned 2d ago · by lpm-firewall-aiNo confirmed malicious attack surface. The package is a Meetless AI-agent governance CLI with explicit user-command wiring for Claude Code hooks/MCP/skills and authenticated Meetless API access, not install-time mutation.
Decision evidence
public snapshot- package.json has no preinstall/install/postinstall hooks; only bin maps mla to dist/cli.js.
- README.md documents explicit user commands: login, activate, doctor, uninstall; npm install only installs the command.
- dist/lib/wire.js writes Meetless-owned Claude hooks, skills, scout agents, and mcpServers.meetless only via runWire/init/rewire.
- dist/commands/activate.js writes non-secret .meetless.json and may bootstrap an installed session-start hook only after user activation.
- dist/lib/config.js defaults to control.meetless.ai/intel.meetless.ai/app.meetless.ai and stores CLI auth config at 0600.
- dist/connectors/claude-code/plugin-detect.js dynamic subprocess use is local `claude plugin list --json` detection for mla@meetless ownership.
Source & flagged code
4 flagged · loading sourcePackage source references a known benign dynamic code generation pattern.
dist/commands/ask.jsView on unpkg · L100Package source references dynamic require/import behavior.
dist/connectors/claude-code/plugin-detect.jsView on unpkg · L58Package ships non-JavaScript build or shell helper files.
dist/hooks-template/user-prompt-submit.shView on unpkgThis package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
dist/commands/activate.jsView on unpkg