#context-management

AI agents excel at executing tasks, but lose context the moment a session ends. Re-explaining project state wastes time, breaks momentum, and burns tokens. **CWIM Kanban bridges the gap.** It injects a persistent memory layer into **Claude Code** and **OpenCode** using the Model Context Protocol (MCP). * 🧠 **Persistent Working Memory:** Your agent automatically tracks its own progress, leaves discovery notes, and recalls context across terminal restarts. * 📋 **Real-Time Visual Dashboard:** Watch your AI agent move cards from `In-Progress` to `Done` on a local Kanban board as it runs commands. * 📂 **Git-Driven Session Isolation:** Auto-detects your repository. When you `cd` into a project, your AI's memory switches context instantly.

One-line MCP config gives your AI agent persistent memory with automatic workflow rules. Unlike passive memory stores, Awareness teaches your agent WHEN to remember and recall — works across Claude Code, Cursor, Windsurf, Cline, Copilot and 7 more IDEs.

Hierarchical memory for AI agents. Three-layer (buffer/working/core) with automatic decay, promotion, and semantic search.

RLM solves Claude Code's biggest limitation: memory loss. Every time you /compact or hit the context limit, your decisions, insights, and conversation history disappear. RLM fixes this with an MCP server that auto-saves before context loss, lets you store key decisions as searchable insights, and keeps your full conversation history in persistent chunks. Features include BM25 ranked search, fuzzy matching (typo-tolerant), multi-project organization, and smart retention that auto-archives old chunks. Inspired by MIT CSAIL's Recursive Language Models paper. Install in 3 lines, zero configuration needed.

Discover the power of the “CLGateway”, an the open-source Claude MCP Server Gateway (available on GitHub). Built for users of Claude Desktop, Claude API or Claude Code, this Python-based gateway smartly manages your MCP (Model Context Protocol) servers and dramatically slashes token usage, cutting MCP context bloat by roughly 95%. Rather than forcing Claude to load every tool from all configured MCP servers (often thousands of tokens wasted upfront), the gateway exposes just three core tools initially. When you need a specific MCP server (GitHub, Google Workspace, DataForSEO, etc.), the gateway dynamically loads only what’s required, reducing tokens used, improving context clarity, and boosting performance.

MCP server code for context-mcp

Personal Context Technology for AI Personalization MCP server

Mirror of

Multi-project memory MCP - A Model Context Protocol server for managing project memory and context