The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to use standard shell commands, specifically grep for retrieval and echo for storage. These operations are scoped to specific directories (.agents-workshop/braindump/ or ~/.braindump/) to prevent arbitrary file system access.
[INDIRECT_PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks common to long-term memory systems, where previously stored facts could contain instructions designed to influence future agent behavior.
Ingestion points: The agent retrieves information by executing grep on markdown files in .agents-workshop/braindump/ and ~/.braindump/ (SKILL.md).
Boundary markers: There are no explicit delimiters or boundary markers recommended for the storage or retrieval process to separate data from instructions.
Capability inventory: The skill enables the agent to perform file writes via echo and file reads via grep commands.
Sanitization: The skill does not include instructions for sanitizing or filtering content before it is read into the agent's context.

braindump