Hivemind
Give your coding agent stable, local-first access to an Obsidian vault so specs and notes stay the single source of truth.
Overview
Hivemind is an MCP server for the Build phase that connects AI agents to a local Obsidian vault for consistent, canonical context.
What is this MCP server?
- Local-first MCP server for Obsidian vaults (@hiveforge/hivemind-mcp)
- stdio transport with HIVEMIND_VAULT_PATH required and optional HIVEMIND_CONFIG_PATH
- Keeps canonical context consistent across AI tool sessions
- No cloud vault lock-in—files stay on your machine
- Version 3.6.3 published on npm for MCP-compatible agents
- Package @hiveforge/hivemind-mcp version 3.6.3
- stdio MCP transport
- Required env: HIVEMIND_VAULT_PATH; optional: HIVEMIND_CONFIG_PATH
Community signal: 7 GitHub stars.
What problem does it solve?
Agents forget yesterday's vault notes unless you re-paste them, so solo builders lose single-source-of-truth context across sessions.
Who is it for?
Solo builders who already use Obsidian as a PKM or spec store and want MCP access without moving knowledge to SaaS.
Skip if: Teams standardized on Notion or Confluence only, or anyone unwilling to expose a local vault path to an MCP client.
What do I get? / Deliverables
Your agent reads and works against the same Obsidian vault over MCP so specs and notes stay aligned while you build and document.
- MCP-accessible Obsidian vault with consistent agent context
- Optional Hivemind config-driven behavior via HIVEMIND_CONFIG_PATH
- Local-first note retrieval without migrating docs to a cloud KB
Recommended MCP Servers
Journey fit
Canonical shelf is Build because the server exists to connect agents to your knowledge base while you implement and document. Agent-tooling captures MCP bridge behavior—vault path, optional config—not a standalone diagram or deploy skill.
How it compares
Obsidian vault MCP bridge, not a diagram generator or a Japan market data API.
Common Questions / FAQ
Who is Hivemind for?
Indie developers and solo builders who keep product and technical knowledge in Obsidian and use Claude Code, Cursor, or similar MCP agents.
When should I use Hivemind?
Use it during build and docs work—and ongoing iteration—whenever the agent should see the same vault-backed context every session.
How do I add Hivemind to my agent?
Install @hiveforge/hivemind-mcp, set HIVEMIND_VAULT_PATH to your vault folder, optionally set HIVEMIND_CONFIG_PATH, and add the stdio server to your MCP configuration.