The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a bundled Python script (normalize_transcript.py) to parse and normalize transcript data into structured JSON format. This script is restricted to the Python standard library, performs only local text processing, and does not perform any network operations or access sensitive file paths.- [PROMPT_INJECTION]: The skill processes untrusted transcript data, creating an indirect prompt injection surface. This risk is effectively mitigated by the skill's design. Ingestion occurs via the normalization script in SKILL.md. Boundary markers are established through the use of structured JSON segments. Capability inventory includes drafting follow-up emails or messages. Sanitization is enforced through explicit instructions requiring the agent to obtain human review and confirmation before sending any drafted content.- [SAFE]: No malicious patterns, obfuscation, or unauthorized data access were detected. All external resource patterns and CLI operations align with the legitimate intended functionality of the skill.

meeting-recap