The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data to populate legal document templates, which introduces a theoretical surface for indirect prompt injection where a user might attempt to embed instructions within contract terms.
Ingestion points: User-provided context such as names, Thai ID numbers, addresses, and specific contract conditions are collected through the /lawyer-pro command as described in SKILL.md and templates/prompt-main.md.
Boundary markers: The instructions lack specific boundary markers or explicit directions for the agent to treat user-provided context as inert data or to ignore any potential instructions embedded within that context.
Capability inventory: The skill's functionality is limited to generating Markdown text documents based on templates; it does not utilize tools for network operations, file system modification, or shell command execution.
Sanitization: There is no explicit sanitization or validation logic for user-provided data before it is interpolated into the final document templates. Given the restricted output capability, this surface is considered a functional requirement rather than a significant security risk.

lawyer-pro