The Agent Skills Directory

[SAFE]: The skill implements proactive security measures against CSV/Formula Injection (also known as Formula Injection or CSV Injection). It explicitly instructs the agent to prepend an apostrophe (') to any cell values starting with sensitive characters (=, +, -, @, etc.). This prevents malicious strings in legal documents or evidence from being executed as formulas when a user opens the resulting CSV or Excel files.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes untrusted external data, such as patent specifications, prior art references, and deposition transcripts.
Ingestion points: Document mapping involves reading external evidence corpus, including produced documents and expert reports.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" warnings for the ingested data.
Capability inventory: The skill has the ability to write files (Markdown, CSV, Excel) but lacks network access or shell execution capabilities.
Sanitization: The skill includes sanitization logic for spreadsheet outputs to prevent formula execution.
[DATA_EXPOSURE]: The skill accesses local configuration files and matter logs located in ~/.claude/plugins/config/claude-for-legal/. These paths are specific to the tool's own environment and are used to maintain role, work-product headers, and matter-specific context as part of its intended legal practice management functionality.

claim-chart