skills/vllnt/skills/explain-subject/Gen Agent Trust Hub

explain-subject

Pass

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external data to generate explanations, creating a potential vector for malicious instructions embedded in those sources.
  • Ingestion points: The workflow explicitly requires the agent to inspect "relevant code or authoritative sources" (SKILL.md, Workflow Step 2) and process a user-provided "real subject."
  • Boundary markers: The skill provides logical boundaries (e.g., distinguishing facts from assumptions) but lacks technical delimiters or markers to isolate untrusted external content from the agent's core instructions.
  • Capability inventory: The referenced Quality Validation protocol (quality-validation.md) describes procedures for "authorized code-change mode," including comparing task deltas and justification of code additions. This indicates the agent environment may support file modification or execution capabilities.
  • Sanitization: The instructions do not define methods for sanitizing, filtering, or escaping content retrieved from external sources before it is interpreted by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 5, 2026, 08:18 PM