customer-interviewing

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill package contains a development utility script evals/run-static-checks.sh which executes shell commands and interprets an embedded Python block to perform consistency and structural checks on the skill's local files.
  • [PROMPT_INJECTION]: The skill's primary function involves analyzing untrusted user content such as interview notes and transcripts, which presents a surface for indirect prompt injection.
  • Ingestion points: User-provided materials are ingested and processed by the critique-questions, conduct, and synthesize intents as specified in SKILL.md.
  • Boundary markers: There are no explicit instructions or delimiters defined in the workflow to isolate user content or warn the agent against executing embedded instructions.
  • Capability inventory: The skill logic focuses on text synthesis and transformation using local markdown playbooks and CSV routing; it does not utilize network tools or shell access during normal agent operations.
  • Sanitization: No sanitization or validation of the ingested user material is performed prior to analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 09:38 PM
Security Audit — agent-trust-hub — customer-interviewing