Cachetank Mcp
Sync personal context into every MCP-capable AI tool so sessions resume with your preferences, facts, and prior decisions.
Overview
CacheTank MCP is an MCP server for the Build phase that loads and optionally writes your CacheTank personal context into AI coding tools over stdio.
What is this MCP server?
- CacheTank MCP supplies shared personal context so tools do not cold-start each chat
- Read access via required CACHETANK_READ_TOKEN from cachetank.com/account
- Optional CACHETANK_WRITE_TOKEN enables the cache_it write tool
- stdio npm package cachetank-mcp at version 2.1.2
- Designed for multi-tool solo workflows across Claude Code, Cursor, and similar clients
- Package version 2.1.2 (cachetank-mcp)
- Two environment variables: required read token, optional write token
- Transport: stdio; registryType npm
Community signal: 1 GitHub stars.
What problem does it solve?
Switching between AI tools forces you to repeat background context and lose thread on long solo builds.
Who is it for?
Solo builders using multiple MCP clients who want one hosted personal context layer without running their own memory server.
Skip if: Builders who require fully offline, on-disk-only memory or cannot use CacheTank cloud tokens.
What do I get? / Deliverables
Each MCP session can read shared CacheTank context—and write updates when configured—so agents start from your account state instead of zero.
- MCP tools that read CacheTank personal context into the active agent
- Optional write path to persist new context via cache_it
- Shared context layer across registered MCP clients
Recommended MCP Servers
Journey fit
Context persistence sits in Build/agent-tooling because you wire the memory layer while composing agents, not when you are only validating a landing page. Agent-tooling is the shelf for cross-client MCP bridges that make Claude, Cursor, and Codex share one personal knowledge base.
How it compares
Hosted personal-context bridge over MCP—not a local markdown memory stack like sqlite-vec on your machine.
Common Questions / FAQ
Who is CacheTank MCP for?
Solo developers who use several AI tools and want one CacheTank profile to prime every MCP session.
When should I use CacheTank MCP?
Use it while building agent workflows when you are tired of re-uploading context and want read/write personal memory from the agent.
How do I add CacheTank MCP to my agent?
Add the cachetank-mcp stdio server in your MCP config, set CACHETANK_READ_TOKEN, add CACHETANK_WRITE_TOKEN if you need cache_it writes, then restart the client.