The Agent Skills Directory

Local File Access: The skill is designed to read specific local markdown and YAML files within the ~/.claude/plugins/config/claude-for-legal/ directory. This is a functional requirement for retrieving matter history and risk data to generate the requested briefing. The skill includes a 'Conflicts gate' logic that validates the requested matter identifier against a log file before proceeding, which serves as a control to ensure only managed data is accessed.
Indirect Prompt Injection Surface: The skill processes external data (matter summaries and history logs) to produce its output. While this presents a theoretical surface for indirect prompt injection if those files were to contain malicious instructions, the risk is mitigated as the skill lacks capabilities to perform sensitive operations such as network requests, file modifications, or shell command execution. The impact of such an injection would be limited to the generated text of the briefing itself.
Data Handling: The skill processes sensitive legal information including risk ratings, materiality, and exposure. This data is handled locally to generate a briefing for the user and is not transmitted to external services, aligning with the expected behavior of a legal management tool.

matter-briefing