PowerMem

PowerContext, the successor to PowerMem, keeps the context of a task with the task itself so that a person or an agent can resume work someone else started. Instead of leaving reasoning and current state behind in a single conversation, it stores what the work needs now — confirmed decisions, constraints, progress, evidence, and next steps — and makes that available to whoever or whatever picks the task up.

You decide what has to move with the task. Durable information is stored as Memory, the current objective and state are organised into a Handoff, and reusable approaches can be recorded as Experience or Skill. Every item stays scoped to its piece of work and keeps its sources and earlier revisions.

A local server holds the context, using a SQLite database by default, and a command-line tool sets up per-agent integrations. Supported agent clients and IDEs connect through the Model Context Protocol (MCP) or a dedicated integration, and applications reach the same store through an async Python client, an HTTP API, MCP, or an in-process Core SDK. An optional personal service keeps the server running after the terminal closes and can start again at login.

Features

• Memory: durable facts about the work, kept with their sources and earlier revisions
• Handoff: the current objective and state, packaged so another person or agent can continue
• Experience and Skill: reusable approaches recorded alongside the work they came from
• Scoped storage: every item stays within the scope of the work it belongs to
• Agent integrations: setup commands wire up coding agents and IDEs, pinned to a matching release ref
• Entry points: async Python client, HTTP API, MCP, or the in-process Core SDK
• Local server: context is stored in a local SQLite database by default
• Personal service: an optional background service survives terminal closure and can start at login