The Agent Skills Directory

[SAFE]: The skill's primary function is to organize and document session progress. It implements user-confirmation steps for destructive file operations (overwriting or appending), which aligns with security best practices for file management.
[DATA_EXPOSURE]: The skill analyzes the current conversation history to generate summaries. It does not attempt to access sensitive system directories, credentials, or perform unauthorized network requests.
[INDIRECT_PROMPT_INJECTION]: As the skill processes untrusted user input from the session transcript to generate summaries, there is an inherent surface for indirect prompt injection. However, the scope is restricted to generating markdown text at a user-specified path, and the instructions emphasize manual oversight for destructive actions, reducing the risk of malicious instruction propagation. Evidence: Ingestion points (session transcript), Boundary markers (none), Capability inventory (file listing, file writing, user interaction), Sanitization (none).

handoff-doc