The Agent Skills Directory

[COMMAND_EXECUTION]: The entry point brain.cjs executes secondary scripts using the Node.js child_process.spawn method. This is used to route user commands to the appropriate internal logic files. The execution is confined to internal scripts and arguments are not processed through a shell.
[PROMPT_INJECTION]: The skill provides an indirect prompt injection surface by design. It stores untrusted data that is later loaded into the agent's context during retrieval.
Ingestion points: scripts/remember.cjs and scripts/lib.cjs save user-provided text into local Markdown files in the ~/.learnwy/ai/memory/ directory.
Boundary markers: The skill does not use delimiters or instructions to isolate recalled memories when they are presented to the agent in scripts/start.cjs or scripts/recall.cjs.
Capability inventory: The skill possesses file system capabilities (read, write, and delete) restricted to its local storage directory.
Sanitization: User-provided content is stored and retrieved as plain text without validation, escaping, or integrity checks.

ai-brain