explorer-remote-proxy
Fail
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: HIGHDATA_EXFILTRATIONREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [DATA_EXFILTRATION]: The skill instructions direct the agent to search through multiple sensitive shell configuration files, including
~/.zshrc,~/.zshenv,~/.zprofile,~/.profile, and~/.bashrc. These files are standard locations for storing plain-text secrets, API keys, and other sensitive environment variables. Accessing these files exposes their full content to the agent's context. - [REMOTE_CODE_EXECUTION]: An automated scan flagged a
curl | python3pattern as a potential remote code execution risk. Manual inspection of Step 3 shows the commandcurl -s "http://localhost:8080/..." | python3 -c '...'uses Python's-cflag with a static inline script to parse JSON data. It does not execute the dynamic content returned by the network request as code, and thus the execution risk is mitigated. - [COMMAND_EXECUTION]: The skill makes extensive use of Docker Compose commands (
docker compose up,docker compose build) andcurlfor interacting with local and remote endpoints. These commands are consistent with the skill's stated purpose of managing a containerized proxy environment.
Recommendations
- HIGH: Downloads and executes remote code from: http://localhost:8080/dependency_graph/graph?root_node_url=$ENC&limit=200 - DO NOT USE without thorough review
- AI detected serious security threats
Audit Metadata