ad-script-generator

Pass

Audited by Gen Agent Trust Hub on Apr 18, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted competitor ad data.
  • Ingestion points: Competitive ad intel dump from the ad-research skill (Step 1).
  • Boundary markers: Absent. The instructions do not define delimiters or warnings to ignore instructions within the ingested text.
  • Capability inventory: Use of filesystem for reading context and writing scripts, sequential MCP for multi-step reasoning, and instructions for external API routing.
  • Sanitization: Absent. The process extracts content (hook, body copy) without filtering for adversarial instructions.
  • [DATA_EXFILTRATION]: The skill accesses sensitive local business data which could be exposed if the agent is compromised.
  • The skill reads configuration files from ~/.claude/contexts/growth/, including customer profiles (icps/), product facts (products/), and internal writing rules.
  • Step 5 (Production Routing) instructs the agent to send data to "Arcads (or equivalent) via API," which implies a network operation involving the generated scripts and potentially internal context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 18, 2026, 12:07 PM
Security Audit — agent-trust-hub — ad-script-generator