telemetry-insight-engine

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill is designed to ingest and analyze external telemetry data, which is an untrusted source.
  • Ingestion points: The input argument in SKILL.md accepts paths to telemetry data (JSON).
  • Boundary markers: No delimiters or instructions to ignore embedded prompts within the JSON data are present.
  • Capability inventory: The skill is designed to feed requirements into a secondary tool, the requirements-wizard, creating a multi-step chain vulnerability.
  • Sanitization: No sanitization or validation logic is specified for the telemetry data input.
  • Data Exposure (LOW): The skill documentation explicitly mentions the integration of 'Confidential' and 'Personal' knowledge tiers including 'secrets'. While intended for legitimate analysis, this confirms the skill has access to sensitive data surfaces.
  • Metadata Poisoning (SAFE): The skill includes an authoritative safety claim ('ensuring no leaks to public outputs') in its Knowledge Protocol. Per security guidelines, this is treated as an unverified claim rather than a security guarantee.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:38 PM