The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using grep and sed to manage file paths and search for duplicate entries in the local memory file (MEMORY.md).
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input via the /si:remember command and appends it to a persistent memory file that the agent reads at the start of every session.
Ingestion points: User-provided strings passed to the /si:remember command in SKILL.md.
Boundary markers: Minimal; content is stored as simple markdown list items without escaping or explicit instruction delimiters.
Capability inventory: The agent has capabilities to read/write files and execute shell commands.
Sanitization: No sanitization, validation, or filtering is performed on the stored text, allowing a user to persist malicious instructions that the agent might follow in subsequent sessions.

remember