helm

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill requires the use of WebSearch to collect market data and benchmarks, which exposes the agent to untrusted external content that could contain malicious instructions.
  • Ingestion points: The skill ingests external data during the SURVEY phase (SKILL.md) and through mandatory web search triggers (SKILL.md).
  • Boundary markers: The skill explicitly references _common/WEB_FETCH_SAFETY.md for prompt-injection checks on data pulled via WebFetch or WebSearch.
  • Capability inventory: Across all referenced files (SKILL.md and files in reference/), the skill is restricted to analysis and simulation; it explicitly excludes code execution, executive decision-making, and file system modifications.
  • Sanitization: Relies on the ecosystem-wide prompt-injection check for all external data fetches as defined in _common/WEB_FETCH_SAFETY.md.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 06:56 AM
Security Audit — agent-trust-hub — helm