The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes meeting notes and transcripts, which are external data sources that could contain malicious instructions.
Ingestion points: SKILL.md identifies that transcripts and summaries are fetched via the granola-cli tool.
Boundary markers: There are no instructions provided to the agent to treat the retrieved content as data only or to use specific delimiters to prevent command execution from within the data.
Capability inventory: The skill is authorized to execute commands via the granola-cli binary as specified in the manifest.yaml.
Sanitization: The documentation does not describe any sanitization or validation of the meeting content before it is processed by the agent.

granola