pitch-narrative-builder
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted external data such as scraped web content, reviews, and exports, creating a surface for indirect prompt injection attacks.
- Ingestion points: Data sources include connectors.md (e.g., Firecrawl, Google Search Console) and user-provided project files.
- Boundary markers: The SKILL.md and references/skill-contract.md files explicitly instruct the agent to treat such data as untrusted evidence and ignore embedded instructions.
- Capability inventory: The skill can read shared product marketing context and interact with external marketing platforms via connectors.
- Sanitization: Directs the agent to distinguish direct observations from assumptions and verify claims.
- [EXTERNAL_DOWNLOADS]: The skill references various external tools and platforms for data enrichment and marketing execution.
- Evidence: references/connectors.md lists integrations with well-known services like Google Ads, LinkedIn, HubSpot, and GitHub. These references are for standard functionality and do not include unverified third-party scripts.
Audit Metadata