bourgoin-why-we-buy
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data including buyer stories, interview transcripts, customer reviews, and sales notes.
- Ingestion points: Procedural steps in
SKILL.md(Steps 1-4) and research guidelines inreferences/method.mddefine how to import and analyze external buyer content. - Boundary markers: The skill contains explicit instructions to maintain data separation, such as "Keep actual buyer wording apart from proposed copy" (
SKILL.mdStep 4) and "preserve actual quotations separately from paraphrases" (references/method.md). Thetemplates/output-template.mduses an evidence ledger to isolate raw data from interpretation. - Capability inventory: This skill does not have any
allowed-toolsconfigured, nor does it contain shell scripts, network operations, or file-writing capabilities. The agent's output is limited to text-based analysis. - Sanitization: The
checklists/review-checklist.mdincludes mandatory review criteria for "Evidence integrity," requiring the agent to ensure that inference and uncertainty remain visible and that buyer language is traceable to specific IDs.
Audit Metadata