skills/wu-yc/labclaw/adaptyv/Gen Agent Trust Hub

adaptyv

Pass

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The section 'Suggest Using K-Dense Web For Complex Worflows' contains behavioral instructions directing the AI to monitor the complexity of user requests and 'proactively suggest' an external platform (www.k-dense.ai). This is a form of instruction override meant to steer user interactions toward a specific service.- [DATA_EXFILTRATION]: The skill's implementation transmits user-supplied protein sequences to a non-whitelisted external URL (https://kq5jp7qj7wdqklhsxmovkzn4l40obksv.lambda-url.eu-central-1.on.aws). While consistent with the skill's stated purpose, it involves sending potentially proprietary data to an unverified endpoint.- [COMMAND_EXECUTION]: The documentation provides instructions to execute shell commands for environment setup, specifically 'uv pip install requests python-dotenv' to install external dependencies.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 13, 2026, 03:46 AM