ti-ingest

Pass

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches vulnerability enrichment data (score, severity, CWE) from the NIST National Vulnerability Database using the nvd-lookup.py tool.
  • [COMMAND_EXECUTION]: Executes local utility scripts tools/ti-ingest.py and tools/nvd-lookup.py to process signal payloads and perform enrichment lookups.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface through the ingestion of untrusted threat intelligence signals.
  • Ingestion points: The skill reads signal.json files containing a claim field, which is defined as a free-text description of a vulnerability.
  • Boundary markers: No boundary markers or instructions to ignore embedded content are used when processing the claim data.
  • Capability inventory: The skill writes processed data to the queue/ directory. This data is subsequently consumed by the Exploitability Validation Run task, creating a path for injected instructions to reach downstream components.
  • Sanitization: The skill lacks evidence of sanitization or filtering for the user-provided claim field before it is incorporated into the queued engagement-scope rows.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 13, 2026, 11:34 PM
Security Audit — agent-trust-hub — ti-ingest