story-bank-builder
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from external, potentially untrusted sources which could contain hidden malicious instructions.\n
- Ingestion points: Untrusted data enters the agent context via scraped web text, reviews, comments, and emails as described in
SKILL.mdandreferences/connectors.md.\n - Boundary markers: Explicit instructions are present in
SKILL.mdandreferences/skill-contract.mdthat command the agent to treat external content as untrusted evidence and specifically not to follow any embedded instructions.\n - Capability inventory: The skill defines operations in
SKILL.mdfor reading context and writing artifacts to the response.references/connectors.mddescribes tool layers for search and marketing execution, which are instructed to default to read-only behavior.\n - Sanitization: Absent; the skill relies on high-level behavioral constraints rather than programmatic filtering or escaping of input data.\n- [EXTERNAL_DOWNLOADS]: The
references/connectors.mdfile identifies several well-known technology services and tools, such as Google Search Console, Stripe, HubSpot, and LinkedIn, as potential data sources. These are documented neutrally as part of the skill's intended connectivity layer.
Audit Metadata