supervisor-review-queue
Pass
Audited by Gen Agent Trust Hub on Jul 12, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill manages a review queue that ingests and displays legal drafts and student notes, creating an attack surface for indirect prompt injection.
- Ingestion points: Data is loaded from
references/review-queue.yaml, which includes drafts, intake forms, memos, and client letters. - Boundary markers: The instructions lack explicit delimiters or instructions to the agent to disregard commands within the ingested content during the review process.
- Capability inventory: The skill can read file contents, display them to the user, and perform file writes to update the queue status and logs.
- Sanitization: No specific sanitization or escaping mechanisms are described for the student-provided content before it is processed by the agent.
- [DATA_EXFILTRATION]: The skill accesses a configuration file located in the user's home directory under a hidden application folder (
~/.claude/plugins/config/claude-for-legal/legal-clinic/CLAUDE.md). While used for setting operational modes, accessing the agent's internal configuration environment is a sensitive operation.
Audit Metadata