Agent Mailbox Light

This Skill addresses the need for low-noise, temporary coordination messages between multiple AI workspaces, while ensuring these messages persist across sessions. In practical development, developers often need to pass temporary context prompts, warnings, or status updates between different workspaces (e.g., different projects or agents). Such information isn't suitable for long-term memory storage but still needs to be readable in subsequent sessions.

Usage is straightforward: simply run the script in the sender workspace to write the mail file into the target workspace's .agent-mailbox/inbox/ directory. Later, the recipient agent can automatically check its workspace's inbox and keep folders upon startup or before executing tasks, read the latest few messages, compress their content into a brief work summary, and archive or retain the mail as needed.

It is ideal for teams or individuals who need to coordinate multiple parallel AI workspaces or agent projects—particularly those already using frameworks like OpenClaw, who need to pass temporary, suggestive information between different tasks or projects without introducing complex message queues or central registries. It is especially useful for managing complex projects that require synchronizing temporary context across agents.

It is recommended to integrate mailbox checking into the agent's baseline startup process, reading only the latest 1–3 messages each time to optimize token consumption. Note that this is primarily a 'best-effort' local filesystem mailbox; it does not guarantee delivery, lacks push notifications, and is not suitable as an execution bus or command system. Mail content should be treated as advisory prompts for the local agent's judgment.