The Agent Skills Directory

Formula Injection Defense: The skill includes explicit instructions to neutralize formula injection attacks in CSV and spreadsheet exports. By requiring the prefixing of potentially executable characters (like =, +, -, @) with a single quote, it protects users who open the generated data in external tools like Excel or Google Sheets.
Data Ingestion Considerations: The --update --from-report functionality allows the ingestion of external documents, such as PDF, CSV, or Excel files from registered agents. While this creates a surface for indirect prompt injection—where instructions hidden in a document could attempt to influence the agent—the skill is designed to extract specific structured data fields, which limits this risk factor.
Scoped File Operations: The skill reads from and writes to a designated configuration directory (~/.claude/plugins/config/claude-for-legal/). This localized file access is necessary for maintaining the persistence of the compliance tracker and follows standard practices for plugin-based data management.
Consequential Action Gate: A prompt-based warning mechanism is included for sensitive actions like recording formal filings. This ensures a human-in-the-loop check, especially for users identified as non-lawyers, regarding the legal implications of the data being recorded.

entity-compliance