The Agent Skills Directory

[DATA_EXPOSURE]: The skill provides a script to read meeting data from the local application cache at ~/Library/Application Support/Granola/cache-v3.json. This access is necessary for the stated purpose of bulk data export from the local environment.
[COMMAND_EXECUTION]: The skill utilizes Python and curl to automate data exports and archival tasks. The Python script parses ProseMirror-formatted JSON from the local cache and writes it to the user's desktop, while curl is used to interact with the official Granola API (api.granola.ai).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process meeting transcripts and notes. As these are external data sources ingested into the agent's context, there is an inherent attack surface for indirect prompt injection. However, the risk is minimized by the structural nature of the export processing, which treats the data as text nodes rather than executable instructions.

granola-data-handling