The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves importing data from untrusted external sources, which constitutes a significant attack surface for indirect prompt injection.
Ingestion points: Data is retrieved from Gmail messages, calendar event details, Twitter archives, and conversation logs from other AI services (SKILL.md).
Boundary markers: The instructions describe logic for filtering 'noise' and automated notifications but do not specify the use of delimiters or explicit instructions to ignore embedded prompts during the ingestion process.
Capability inventory: The skill has permissions to write to multiple brain directories (people/, conversations/, etc.), modify system timelines, and execute shell commands via the gbrain CLI tool.
Sanitization: There are no mentioned mechanisms for sanitizing or escaping the imported text before it is stored in the knowledge base and subsequently retrieved by the agent.
[COMMAND_EXECUTION]: The skill instructs the agent to run various shell commands for local environment discovery and data management.
Evidence: The agent uses shell snippets containing find, du, and wc to discover markdown repositories, and executes the gbrain CLI for importing files and extracting links.

cold-start