The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of meeting notes and transcripts provided by the user (SKILL.md). Boundary markers are absent, meaning there are no specific instructions to ignore embedded commands. However, a capability inventory shows the skill is restricted to reading a local context file and generating text, with no network access or command execution capabilities. No explicit sanitization is implemented for the processed text, but the limited toolset restricts potential impact to prompt-level manipulation.
[SAFE]: No malicious patterns, command execution, or data exfiltration attempts were detected. The use of a local context file follows standard context-sharing practices for AI agents.

meeting-prep