The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources including Gmail inbox messages, sent mail, and Granola meeting notes to generate email drafts. This creates a surface for indirect prompt injection where malicious content in an email could attempt to influence the agent's behavior during the drafting process.
Ingestion points: Gmail messages and meeting notes read via the personalized_email.py script.
Boundary markers: None explicitly defined in the skill instructions to separate untrusted data from instructions.
Capability inventory: The skill involves read-only access to Gmail, local file reading (credentials, configuration), and printing generated content to the console.
Sanitization: No specific sanitization or filtering of the external data is described in the prompt instructions.
[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts (personalized_email.py) and CLI tools (cognee-cli) using the uv run command. These are vendor-provided resources used for the skill's primary functionality of data ingestion and draft generation.

personalized-email