kta-idea-pipeline
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from various public websites to synthesize app ideas, which represents a potential attack surface for indirect prompt injection.
- Ingestion points: Phase 0 uses the
WebFetchtool to retrieve data from user-generated content platforms listed inreferences/discovery-sources.md(e.g., Reddit, ProductHunt). - Boundary markers: The instructions lack explicit delimiters or instructions to the agent to ignore potential malicious prompts embedded within the fetched web content.
- Capability inventory: The skill has file-writing capabilities (
plans/{slug}/idea-brief.md) and the ability to chain execution to another skill (kta-prd-pipeline). - Sanitization: No explicit sanitization or filtering logic is defined for the external data before it is integrated into the agent's context.
- [EXTERNAL_DOWNLOADS]: Fetches research data and trending information from well-known public services.
- Evidence:
references/discovery-sources.mdcontains URLs for the Google Play Store, ProductHunt, Reddit, and IndieHackers. - Note: These are well-known technology platforms and the data retrieval is aligned with the skill's primary purpose of idea discovery.
Audit Metadata