mailbox

To persist agent memory across sessions, use a durable mailbox that stores per-agent findings, context, and standing orders under a dedicated directory path. It enables read, write, load, archive, and list operations to maintain continuity and handoffs between runs.

Managing standing orders for AI agents requires per-agent storage that survives session terminations. A persistent mailbox enforces dedicated directories to store and load these instructions, ensuring agents retrieve and apply consistent operational directives across executions.

Pre-loading context into agent prompts involves reading previously archived findings and memory from a durable mailbox before spawning the agent. This spawn-ready approach injects historical session context directly into the prompt to improve cross-session continuity.

Yes, you can use shell scripts to manage cross-session context for team workflows. The mailbox functionality is driven by a shell script that handles memory management operations, enabling team handoffs and per-agent data isolation without requiring external dependencies.

Append-only logging for agent findings works by continuously adding new context entries to a durable storage file without overwriting previous records. This creates an immutable historical trail that supports accurate session archiving and team handoffs.

No, persistent agent memory does not require external databases or dependencies. The mailbox enforces per-agent storage directly within the local file system using shell scripts, ensuring durable context and findings persistence without complex setup.