The Agent Skills Directory

[SAFE]: The skill utilizes a local port scanning mechanism (ports 9223-9232) for auto-discovery of agent instances. This is an internal communication pattern designed for multi-process orchestration within the local environment and does not involve unauthorized network access.
[SAFE]: Orchestration is achieved through a structured message protocol (agent-broadcast and agent-status). While agents ingest data from each other, this is a core requirement of the coordination workflow and is implemented using typed message schemas rather than unsafe execution patterns.
[SAFE]: External references point to the author's own GitHub repository (sarveshsea/memi), which is consistent with the skill's identity and stated purpose.

multi-agent