deposition-prep
Pass
Audited by Gen Agent Trust Hub on May 20, 2026
Risk Level: SAFE
Full Analysis
- Legal Workflow Context: The skill is designed to interact with specialized legal configuration files and eDiscovery platforms to assist in case preparation. This involves reading from local paths (e.g.,
~/.claude/plugins/config/claude-for-legal/) which is standard for specialized agent environments. - Data Ingestion and Indirect Injection Considerations: The skill processes external documents from eDiscovery platforms like Everlaw or Relativity. While these documents constitute untrusted data, the skill includes robust mitigation strategies:
- Verification Markers: It uses explicit markers like
[VERIFY]and[UNCERTAIN]to ensure the lawyer reviews the output for accuracy. - Source Attribution: Every rule and citation is tagged with its source (e.g.,
[Westlaw],[CourtListener]), preventing the silent adoption of potentially misleading content. - Embedded Security Controls: The skill implements a 'Conflicts gate' that requires a matter to be present in a log file before proceeding, serving as a significant administrative control. Additionally, it includes a 'Destination check' to prevent the accidental disclosure of privileged work product to unauthorized parties.
Audit Metadata