list-enrichment

Pass

Audited by Gen Agent Trust Hub on Mar 4, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill manages authentication via the 'EXTRUCT_API_TOKEN' environment variable and provides a terminal command to verify its presence without exposing the full secret value.
  • [SAFE]: Communication is performed exclusively with the vendor's official API at 'https://api.extruct.ai/v1' and documentation at 'https://www.extruct.ai/docs', which are trusted resources for the author 'extruct-ai'.
  • [PROMPT_INJECTION]: The skill contains a vulnerability surface for indirect prompt injection as it interpolates external company data into prompts for research agents. \n
  • Ingestion points: Company domains and table data are retrieved from the user's Extruct tables via 'GET /v1/tables/{table_id}/data' as specified in 'SKILL.md'. \n
  • Boundary markers: The prompt templates do not currently employ specialized delimiters or instructions to ignore embedded commands for the '{input}' variable. \n
  • Capability inventory: The skill can create and trigger 'research_pro' agents which perform web-based research through the 'POST /v1/tables/{table_id}/columns' endpoint. \n
  • Sanitization: No sanitization of the input domain or company data is specified before interpolation.
  • [COMMAND_EXECUTION]: The skill suggests running a local terminal command ('echo $EXTRUCT_API_TOKEN | head -c 5') to verify API credentials; this is a standard safety check for the user to ensure environment variables are correctly set and does not represent a privilege escalation risk.
  • [SAFE]: The skill includes instructions for comparing local API references against live documentation and updating local files, which is standard procedure for maintaining tool compatibility.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 4, 2026, 07:06 PM