The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs shell commands like grep, sed, and pwd to manage memory files located in the user's home directory (~/.claude/projects/). These operations are used for locating, searching, and updating internal memory files.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a data ingestion point, storing user-provided facts and patterns in a MEMORY.md file that the agent reads in future sessions. This creates a surface for instructions to be embedded in data, though the skill includes explicit guidelines to keep entries concise and avoid storing sensitive information like credentials or secrets.
Ingestion points: User input provided to the /si:remember command is written to MEMORY.md.
Boundary markers: The instructions suggest wrapping facts in concise markdown list items, though no explicit sanitization logic for the content itself is provided.
Capability inventory: The skill has the ability to read and append to local files using shell redirection and grep via the agent's bash tool.
Sanitization: The skill relies on the agent's logic to parse and verify knowledge before writing, and it includes a specific 'What NOT to use' section to prevent storing secrets.

remember