Taproot
Persist notes and recall them across Claude and other MCP clients so your agent does not lose project context every new session.
Overview
Taproot is an MCP server for the Build phase that gives AI tools a persistent memory layer to save and recall notes across MCP clients.
What is this MCP server?
- Taproot persistent memory layer for AI tools via remote MCP at connect.taproothq.com/mcp
- Save and recall notes across Claude and other MCP-compatible clients
- Hosted streamable-HTTP transport—version 0.4.0 in official manifest
- Vendor site taproothq.com for account and product context
- 1 remote MCP endpoint (streamable-http)
- Server manifest version 0.4.0
What problem does it solve?
Agents forget prior decisions each session, forcing solo builders to repeat context and ship slower.
Who is it for?
Indie developers using multiple MCP clients who want one shared memory bus for specs, debugging notes, and recurring preferences.
Skip if: Teams that need full document CM in git only, strict on-prem memory with no hosted vendor, or secrets storage without reviewing Taproot security posture.
What do I get? / Deliverables
After connecting the remote Taproot MCP, your agent can store and retrieve notes so multi-day builds and fixes reuse stable project memory.
- Shared save/recall memory tools for connected MCP agents
- Remote MCP endpoint configuration for team IDE setups
Recommended MCP Servers
Journey fit
Persistent memory MCPs are Build-phase agent tooling because they extend how your stack remembers decisions, snippets, and prefs while you integrate and ship features. Agent-tooling is the canonical shelf for cross-client memory layers that sit beside your editor rather than replacing your database or docs repo.
How it compares
Cross-client memory MCP, not a vector database you self-host or a project-management skill.
Common Questions / FAQ
Who is Taproot for?
Solo builders and small teams using Claude and other MCP tools who want durable notes the agent can fetch later without re-pasting context.
When should I use Taproot?
Use it during Build and Operate when long-running projects need remembered stack decisions, bug writeups, and customer constraints across sessions and clients.
How do I add Taproot to my agent?
Add remote MCP URL https://connect.taproothq.com/mcp with streamable-http in your client settings and complete any Taproot account steps from taproothq.com.