Massive Context Mcp
Feed monorepos and huge doc dumps into agent workflows via chunking, sub-queries, and local Ollama inference when context exceeds normal window limits.
Overview
Massive Context MCP is a MCP server for the Build phase that chunks and sub-queries 10M+ token corpora with local Ollama inference for agent workflows.
What is this MCP server?
- Designed for 10M+ token corpora with chunking and sub-query decomposition (v3.0.1)
- Local Ollama inference through OLLAMA_URL (default http://localhost:11434)
- Persistent context workspace under RLM_DATA_DIR
- stdio MCP via PyPI massive-context-mcp or MCPB bundle
- Splits giant inputs so Claude Code, Cursor, or Codex agents query slices instead of truncating silently
- Server version 3.0.1
- Documented scale: 10M+ token contexts
- Default OLLAMA_URL http://localhost:11434
What problem does it solve?
Your codebase and docs exceed what any single agent context window can hold, so answers drop whole subsystems you never knew were missing.
Who is it for?
Solo builders analyzing huge local trees or log dumps with Ollama who need MCP-native chunking rather than a hosted vector DB project.
Skip if: Teams that only work in small repos, need multi-user cloud RAG with ACLs, or cannot run Ollama locally for sub-query passes.
What do I get? / Deliverables
After install, oversized inputs land in RLM_DATA_DIR-backed chunks your agent queries through sub-requests instead of losing the tail of the archive.
- Chunked storage and retrieval for very large text corpora
- Sub-query workflow over local Ollama
- stdio MCP integration at version 3.0.1
Recommended MCP Servers
Journey fit
Large-context handling is core agent-tooling during build when you integrate repos, specs, and logs the model cannot load wholecloth. Chunking plus sub-query orchestration keeps solo builders unblocked on 10M+ token corpora without paying for oversized cloud context on every turn.
How it compares
Local mega-context chunking MCP, not a managed embeddings SaaS or a single-shot summarize skill.
Common Questions / FAQ
Who is Massive Context MCP for?
Indie developers using AI coding agents who regularly work with monorepos, massive documentation, or logs that blow past normal context limits.
When should I use Massive Context MCP?
Use it during build when onboarding the agent to a large codebase or doc corpus and you need structured chunking plus sub-queries instead of manual file picking.
How do I add Massive Context MCP to my agent?
Install massive-context-mcp from PyPI or the v3.0.1 MCPB release, set RLM_DATA_DIR and OLLAMA_URL, then register the stdio server in your MCP configuration.