tiangong-auto-research-workbuddy

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The shell script scripts/workbuddy_research_cli.sh performs a search for a valid Node.js 24 executable across standard installation paths (e.g., NVM, FNM, Homebrew). Once a valid binary is found, it uses exec to run a local script, research_cli.mjs, located in a sibling skill directory. This is a standard method for ensuring environment compatibility and delegating tasks to a specialized tool from the same vendor.
  • [DATA_EXPOSURE]: The script accesses environment variables such as HOME and AUTO_RESEARCH_NODE to resolve file paths for Node.js installations. This behavior is restricted to local configuration discovery and does not involve the exfiltration of sensitive information or unauthorized access to user data.
  • [PROMPT_INJECTION]: The instructions in SKILL.md are safety-oriented, explicitly directing the agent to operate within the 'Default Permission' mode and warning against the use of unsafe flags like dangerouslyDisableSandbox or Full Access. These instructions reinforce the host application's security boundaries rather than attempting to bypass them.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 04:13 AM
Security Audit — agent-trust-hub — tiangong-auto-research-workbuddy