ipt

Pass

Audited by Gen Agent Trust Hub on Oct 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external data from remote IPT instances, including resource titles, metadata, and server log files. Ingestion points: The ipt_health.py script fetches data from user-defined URLs and the official GBIF API (file: scripts/ipt_health.py). Boundary markers: The results are presented in structured tables for interpretation. Capability inventory: The skill can trigger resource publication on remote servers via the publish command. Sanitization: The script applies standard string processing and the skill body includes explicit instructions for human oversight before execution of mutating commands.
  • [SAFE]: The skill follows secure credential management practices and includes protections for sensitive data. It uses system environment variables (IPT_USER, IPT_PASSWORD) to handle authentication securely and documents specific server files as restricted, warning against their disclosure in references/administration.md. All network operations target well-known service endpoints or the user-provided application URL.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 2, 2026, 07:26 AM