reg-gap-analysis
Pass
Audited by Gen Agent Trust Hub on May 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to Indirect Prompt Injection because it processes untrusted regulatory text provided by the user.
- Ingestion points: The skill ingests regulatory text, summaries, or guidelines directly from user input (argument-hint and Step 2 of the workflow).
- Boundary markers: The skill lacks explicit boundary markers or instructions for the agent to ignore embedded commands within the provided regulation text.
- Capability inventory: The skill reads local files (
~/.claude/plugins/config/claude-for-legal/privacy-legal/CLAUDE.md) and references the use of external retrieval connectors (MCP), web search, and file creation. - Sanitization: There is no evidence of input validation or sanitization of the user-provided text before it is used to drive the analysis workflow.
- [DATA_EXFILTRATION]: The skill explicitly reads from a specific local path:
~/.claude/plugins/config/claude-for-legal/privacy-legal/CLAUDE.md. - Finding: While this access is the stated purpose of the skill (to establish current compliance status), the file contains sensitive information such as privacy commitments, regulatory coverage, and internal system inventories (DSAR system list). In a scenario involving successful prompt injection, this data could be targeted for exfiltration via the mentioned search connectors or web search capabilities.
Audit Metadata