daily-ad-check

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from ad account exports, campaign names, and notes, creating an attack surface for indirect prompt injection.
  • Ingestion points: Data is ingested via pasted exports, CSV files, or connected Ad MCPs as described in SKILL.md.
  • Boundary markers: Robust instructions in SKILL.md and references/agent-security.md mandate treating all external content as data and never as instructions. The agent is explicitly told to report embedded instructions as findings rather than complying with them.
  • Capability inventory: The skill is restricted to read-only operations on ad accounts. It writes state snapshots and product context to the local .agents/ directory as specified in SKILL.md and references/run-state.md.
  • Sanitization: Security guidelines in references/agent-security.md require HTML-escaping for any interpolated values in emitted markup to prevent potential cross-site scripting (XSS) when generating code or snippets.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 02:23 PM