deal-team-summary
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill processes diligence findings and data room content, which are external and potentially untrusted sources. This creates a surface where embedded instructions in those documents could attempt to influence the agent's briefing or subsequent deal-related tasks.
- Ingestion points: Reads diligence-issue-extraction output, matter.md, and deal-context.md files from the local filesystem.
- Boundary markers: The skill does not explicitly specify delimiters or instructional warnings to prevent the agent from following directions potentially embedded within the findings.
- Capability inventory: The skill possesses file-write capabilities, specifically writing briefings to the matter folder located at
~/.claude/plugins/config/claude-for-legal/corporate-legal/matters/. - Sanitization: There is no evidence of explicit sanitization or validation of the ingested legal findings before they are interpolated into the briefing templates.
- Local File System Interaction: The skill reads from and writes to specific configuration and matter directories. While these operations appear scoped to the intended plugin environment, they represent an access pattern that should be monitored for least-privilege compliance.
Audit Metadata