The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/granola.py executes the sops utility via subprocess.run to decrypt the Personal API key from a specified file path. This is a legitimate use of a security tool for credential management.
[COMMAND_EXECUTION]: The shell script scripts/sync.sh uses standard command-line tools such as find, grep, sed, and xargs to manage and verify already exported meeting files within the local vault.
[DYNAMIC_EXECUTION]: The scripts/sync.sh script includes an integration point that executes an optional post-sync script (session_loop_postsync.sh) if it is present and executable in the user's vault directory. This allows for workflow extension but involves executing a local script based on its file system path.
[INDIRECT_PROMPT_INJECTION]: The skill ingests meeting transcript and summary data from the Granola API, which is then processed and written to markdown files. This creates a surface for indirect prompt injection if an agent subsequently reads these files and follows instructions embedded within the transcript content.
Ingestion points: Meeting data is fetched from the Granola API (public-api.granola.ai) in scripts/granola.py.
Boundary markers: The skill uses standard markdown headers (## Summary, ## Transcript) to separate content types in the exported files, but does not implement specific delimiters or warnings to prevent an agent from following instructions embedded in the meeting text.
Capability inventory: The skill has capabilities to write files to the local file system and execute local commands and scripts.
Sanitization: The skill performs filename slugification to sanitize output paths and uses standard JSON encoding for metadata, but the transcript text itself is not filtered or escaped.

granola