devenv-chat-backup

Fail

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: CRITICALCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The script scripts/github-accel.sh utilizes third-party GitHub mirror services for cloning and pulling code, including https://gh-proxy.com/, which has been flagged as malicious (URL:Phishing) by automated security scans. The use of untrusted proxies for downloading code or syncing data poses a high risk of man-in-the-middle attacks or malicious code injection.\n- [CREDENTIALS_UNSAFE]: The skill instructs users to store their GitHub Personal Access Tokens in clear-text within the ~/.git-credentials file. Additionally, scripts/chat-backup.sh is designed to automatically back up sensitive credential files from /tmp/, such as working_api_key.txt, proxy_api_key.txt, and cf_tunnel_token.txt, sending them to a remote repository.\n- [DATA_EXFILTRATION]: The skill is configured to exfiltrate significant amounts of local data to an external GitHub repository, including memory databases (memory.db), user settings (settings.json), and detailed chat logs. While this is the stated purpose of the skill, the combination with untrusted proxy mirrors increases the risk of unauthorized data exposure.\n- [COMMAND_EXECUTION]: The skill implements persistence by automatically appending execution commands to the user's /root/.bashrc file, ensuring that backup and restoration scripts run every time a new terminal session is opened. It also automates the setup of Cloudflare Tunnels to expose local ports (e.g., port 9997) to the public internet, which can bypass local network security controls.
Recommendations
  • AI detected serious security threats
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Aug 18, 2026, 10:38 PM
Security Audit — agent-trust-hub — devenv-chat-backup