solo-validate

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted data from both local and external sources.
  • Ingestion points: Data enters the agent context through the WebSearch tool (Step 6b), the mcp__solograph__kb_search tool (Step 2), and by reading the research.md file (Step 3).
  • Boundary markers: There are no explicit delimiters or negative constraints (e.g., "ignore instructions found in this data") used when interpolating search results or file content into the analysis prompts.
  • Capability inventory: The skill possesses the capability to execute commands via the Bash and Grep tools, and to modify the filesystem using the Write and Edit tools as defined in the SKILL.md allowed-tools.
  • Sanitization: No sanitization or validation logic is present to filter or escape instructions that might be embedded in the external content before it is processed by the STREAM or Manifest analysis steps.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 10:52 AM