Context Pod

A remote MCP server that gives AI agents persistent memory through vector-based context with certainty scoring. Agents call get_pod_summary at session start, save_event to record new observations, apply_learning when corrected, and update_session at shutdown. Certainty starts at 0.3 and rises with corroboration. The chat lane lets you audit what any agent learned, correct drift, and add context no agent can observe alone. Multi-pod setups let coding agents, workflow agents, and personal assistants share context or stay isolated. Connects via SSE transport at context.nervousmachine.com/mcp. Useful when you want agents to remember across sessions without re-teaching every time, or when multiple agents need to build on shared knowledge.