The Agent Skills Directory

[DATA_EXFILTRATION]: Accesses sensitive hidden configuration directories. The skill instructs the agent to read from ~/.claude/plugins/config/... to retrieve user profiles. Accessing hidden directories within the user's home folder is a sensitive operation as these locations often contain private configuration data or user-specific preferences.
[INDIRECT_PROMPT_INJECTION]: The skill processes instructions from external source files without safety boundaries. Ingestion points: Reads content from litigation-legal/CLAUDE.md and various files within litigation-legal/skills/ at runtime. Boundary markers: No delimiters or explicit warnings are provided to the agent to ignore instructions that might be embedded within these external legal litigation files. Capability inventory: The skill utilizes tools for local file system access, web verification, and document rendering. Sanitization: There is no evidence of sanitization or validation of the content read from the external markdown files before they are processed by the agent.

chinese-legal-litigation