wage-hour-qa
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill is configured to access sensitive local file paths to manage legal case context.
- Evidence: It reads configuration from
~/.claude/plugins/config/claude-for-legal/employment-legal/CLAUDE.mdand case-specific files namedmatter.mdwithin case-related directories. - Context: While this file access is central to the skill's purpose as a case management assistant, it represents an exposure surface for local data.
- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by aggregating untrusted data from multiple sources.
- Ingestion points: The skill processes user-supplied questions and external search results fetched via MCP tools (yuan dian MCP) or the web.
- Boundary markers: The skill implements source labeling (e.g.,
[yuandian检索],[联网检索——需复核]) which helps differentiate data sources but does not strictly isolate untrusted content from the instruction stream. - Capability inventory: The skill has the capability to read local files, write to case folders, and execute network-based search tools.
- Sanitization: No explicit sanitization or escaping of external content is mentioned before it is processed by the agent.
Audit Metadata