AI Security Review
scanned 6h ago · by lpm-firewall-aiLPM blocks this version under the AI-agent control-surface policy. Install-time code mutates a project-level Opencode skills directory by planting package-supplied agent instructions. This is an unconsented lifecycle write into a broad AI-agent control surface.
Decision evidence
public snapshot- package.json runs postinstall: node skills/postinstall.js
- skills/postinstall.js creates .opencode/skills under process.cwd during install
- postinstall symlinks or copies package .md skills into that AI-agent control surface without opt-in
- skills/browser-automation-opportunities.md promotes automating gemini.google.com as an unlimited API
- No credential harvesting or exfiltration found in JS sources
- No install-time network calls found
- Runtime LLM/image methods are user-invoked and package-aligned
- MCP tools only expose prompt-generation handlers
Source & flagged code
3 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgInstall-time source drops package-supplied AI-agent/MCP control files or instructions.
skills/postinstall.jsView on unpkg · L5