workflow-research

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill coordinates a research workflow that ingests untrusted data from the web, creating a potential surface for indirect prompt injection.\n
  • Ingestion points: The skill invokes $web-research (referenced via ../web-research/SKILL.md) to retrieve content from external web sources.\n
  • Boundary markers: While the instructions establish quality gates for citations and cross-validation, they do not include specific delimiters or system-level instructions to ignore adversarial commands embedded in the researched content.\n
  • Capability inventory: The workflow is authorized to spawn sub-agents and write data to the local file system, specifically in the tmp/reports/ and docs/knowledge/ directories.\n
  • Sanitization: The orchestration logic does not describe any sanitization or filtering mechanisms for the external data before it is processed by synthesis agents.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 10:12 AM