Context Pod
Give Claude Code or Cursor a remote memory layer so long-running agent work keeps scored, vector-learned project context instead of starting cold each chat.
Overview
Context Pod is a MCP server for the Build phase that serves persistent, vector-learned agent context with certainty scores over a remote SSE endpoint.
What is this MCP server?
- Remote SSE MCP at context.nervousmachine.com with Bearer API key auth
- Persistent context built from learned vectors with per-fact certainty scores
- Designed for AI agents that need durable memory beyond a single thread
- Version 1.0.0 server schema with no local npm package in the manifest
- Server version 1.0.0 in MCP registry schema
- Single remote SSE endpoint at context.nervousmachine.com/mcp
- One required secret header: Authorization Bearer API key
Community signal: 2 GitHub stars.
What problem does it solve?
Coding agents lose thread-local memory and cannot rank which past assumptions are still trustworthy across sessions.
Who is it for?
Solo builders running long agent-assisted builds who want hosted persistent memory without self-hosting a vector store.
Skip if: Teams that require fully offline, on-prem-only context with no third-party SaaS endpoint.
What do I get? / Deliverables
After you register the remote MCP server with your API key, agents can pull durable, scored context instead of re-deriving the same project facts from scratch.
- Live remote MCP connection to scored persistent context
- Reusable agent memory across sessions on the same project
- Certainty-aware context facts for tool calls
Recommended MCP Servers
Journey fit
Persistent agent context is installed while you wire up coding agents and integrations, before shipping features that depend on consistent decisions across sessions. Context Pod is agent-tooling: it extends the MCP stack with external memory, not app frontend or deployment plumbing.
How it compares
Hosted MCP memory service, not an in-repo Claude skill or local RAG script.
Common Questions / FAQ
Who is Context Pod for?
Indie and solo developers using MCP-capable agents who need persistent, scored context across many coding sessions on the same product.
When should I use Context Pod?
Use it during Build when you configure agent-tooling and want learned project context to survive beyond a single chat or compaction window.
How do I add Context Pod to my agent?
Add the remote MCP URL https://context.nervousmachine.com/mcp (SSE) in Claude Code, Cursor, or another MCP client and set Authorization to Bearer plus your key from context.nervousmachine.com.