persona-insights-analysis

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt requires quoting transcript verbatims exactly (with speaker attribution) and fetching transcripts/URLs, which could force the model to reproduce any secrets or credentials present in those transcripts verbatim, creating an exfiltration risk.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The skill ingests user-provided sales call transcripts (Source A/B/C/D), and at runtime it will read outsider-authored free text from those transcripts (e.g., prospect/customer dialogue and any transcript text fetched from external URLs), which then becomes LLM context for analysis.