dpa-review
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection Surface: The skill is designed to ingest and analyze untrusted external data in the form of DPAs provided by users through files, links, or text. This represents a potential surface where a maliciously crafted document could attempt to influence the agent's analysis or instructions. \n
- Ingestion points: Untrusted data enters the agent context through the
[file | Drive link | paste text]argument and the subsequent retrieval of the DPA content in the workflow (SKILL.md).\n - Boundary markers: The instructions do not define specific delimiters or instructions to ignore potential commands embedded within the untrusted DPA text.\n
- Capability inventory: The skill possesses the capability to read local configuration and playbook files (~/.claude/plugins/config/...), write outputs to local matter folders, and perform external legal research using network-based tools (SKILL.md).\n
- Sanitization: No explicit sanitization, filtering, or validation of the external DPA content is mentioned before it is processed by the agent.\n- File System Interaction: The skill interacts with the local file system by reading playbook configurations and matter-specific overrides from ~/.claude/plugins/config/claude-for-legal/privacy-legal/ and writing review results to corresponding matter subdirectories. While these paths are specific to the skill's operation, the ability to read and write to the filesystem is a standard security consideration that warrants ensuring access is restricted to the necessary scope.
Audit Metadata