ad-negative-keywords

Pass

Audited by Gen Agent Trust Hub on Sep 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes search term reports, which are untrusted external data sources. This creates a surface for indirect prompt injection where malicious instructions could be hidden in search queries to influence the agent.
  • Ingestion points: Search term reports (SKILL.md, Workflow section).
  • Boundary markers: The instructions do not include explicit delimiters or warnings for the agent to ignore instructions embedded within the search term data.
  • Capability inventory: Data classification, list generation, and script-based tokenization (SKILL.md, Query mining section).
  • Sanitization: No specific validation or sanitization of the input strings is mentioned beyond tokenization for aggregation.
  • [DYNAMIC_EXECUTION]: The skill suggests that the agent can generate and run scripts to tokenize search query exports.
  • Evidence: "Can run scripts: a short tokenizer over the export is the fast path" (SKILL.md, Query mining section).
  • Pattern: Simple script generation for local data processing, which is generally acceptable but noteworthy in the context of processing untrusted inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 13, 2026, 03:32 PM
Security Audit — agent-trust-hub — ad-negative-keywords