A Model Context Protocol server for persistent project-specific AI agent context.
OpenContext MCP gives AI agents persistent, project-local memory.
Most coding agents lose important decisions between sessions: architecture rules, naming conventions, API contracts, migration notes, and hard-won debugging context. OpenContext MCP solves that by exposing a tiny Model Context Protocol server that lets agents save and read markdown files in a local .opencontext/ directory inside each project.
The result is simple and transparent: your agent gets memory, and you keep full control because the memory is plain markdown committed or ignored however you choose.
- savecontext: save markdown context to .opencontext/.md - readcontext: read one saved topic or list all available topics - Project-specific storage based on the client's current working directory - No database, account, cloud sync, or hidden state - Works with MCP-compatible clients over stdio - Built with strict TypeScript and the official @modelcontextprotocol/sdk
You normally do not need to install OpenContext MCP manually. Configure your MCP client to run it with npx:
For local development in this repository:
Add OpenContext MCP to your OpenCode MCP configuration:
Cursor and Claude Desktop use the same MCP server configuration:
Restart your MCP client after updating the configuration.
OpenContext MCP stores files relative to the process working directory used by your MCP client:
Topics must be lowercase snakecase or kebab-case, such as architecture, api-contracts, or codingrules.
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/slxca--opencontext.html)
Memory your agent can trust — every answer carries its source; when it doesn't know, it says so. Hermetic, local, MCP.
naseridev/orvenMCP-compatible context and memory system: paged code navigation for AI assistants, plus a recorded, verified theory of what is known about the code.
azimov777/casefileThe task tracker your AI agents keep for each other: every task carries a case file, so the next agent picks up where the last one stopped. MCP server + live board. Self-hosted, one-line install.
osaxyz/sepiaceLong-term memory for coding agents. The sepiace plugin for Claude Code and Codex.
00200200/usagetrimKeep the signal. Cut the noise. Local CLI + MCP — fold verbose tool output for Claude Code, Codex, Cursor & Desktop.
wahyunuriman999/Muse-Skill-HubOpen-source, LLM-agnostic capability runtime for AI agents — 97 skills as typed, permission-enforced MCP tools. Final certification: 23/23 gates PASS.
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.