agent-reach
Audited by Socket on Sep 20, 2026
2 alerts found:
SecurityAnomalySUSPICIOUS. The skill’s broad internet-research purpose generally matches its capabilities, but its footprint is larger than a simple search skill: it auto-installs/updates executable tooling, encourages browser cookie extraction, and relies on multiple third-party CLIs/MCP servers. Data flows are mostly consistent with the stated purpose, yet the mutable GitHub-archive installer and credential-bearing integrations make this a high supply-chain and credential-handling risk rather than confirmed malware.
The supplied fragment is configuration documentation and does not itself contain executable malicious behavior. It presents significant operational security risks because it requests proxy credentials, API keys, and broad browser-cookie extraction, and it relies on unverified third-party Docker, PyPI, and Git packages. Review the referenced tools before use, restrict MCP services to localhost with authentication where possible, and avoid exporting all browser cookies unless necessary.