The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive local filesystem paths located within a hidden directory in the user's home folder.
Evidence: The skill reads from and writes to files in ~/.claude/plugins/config/claude-for-legal/ip-legal/, including CLAUDE.md and case-specific matter.md files.
Risk: Accessing hidden configuration and data directories exposes internal application settings and potentially confidential legal matter information to the agent's context.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection attack surface by processing untrusted external data.
Ingestion points: The agent is instructed to capture and read "invention disclosures" provided by the user in the form of text or uploaded files in SKILL.md.
Boundary markers: The skill lacks explicit instructions to delimit the user-provided disclosure or to ignore any embedded instructions within that content.
Capability inventory: The skill possesses file read and write capabilities targeting the ~/.claude configuration hierarchy, which could be abused if the agent follows malicious instructions hidden in a disclosure.
Sanitization: No sanitization or validation of the ingested disclosure content is described before it is used to influence the screening process and output creation.

invention-intake