The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to interact extensively with the local mem0-admin CLI. It defines specific workflows for executing commands such as mem0-admin context, mem0-admin review, mem0-admin forget, and mem0-admin dream. It specifically includes a command to apply batch changes (mem0-admin dream --apply "<plan>" --yes) which relies on the AI agent to verify user confirmation before execution, as the CLI-level safety check is bypassed by the flag.
[INDIRECT_PROMPT_INJECTION]: The skill operates on stored memory data, which is considered untrusted input that could contain adversarial instructions targeting the agent's lifecycle or data management logic.
Ingestion points: Memory data is ingested into the agent's context through the output of the mem0-admin review and mem0-admin context commands described in SKILL.md.
Boundary markers: The skill instructions do not provide explicit delimiters or "ignore instructions" wrappers for the retrieved memory content to distinguish it from the agent's primary system instructions.
Capability inventory: The skill has access to sensitive capabilities including data deletion and batch mutation via the mem0-admin CLI and the mem0 MCP tool (delete_memory).
Sanitization: The skill relies on human-in-the-loop verification, requiring the agent to show plans and candidates to the user for explicit approval before any mutating command is executed.

mem0-local-admin