Skip to main content
Glama
July-Blue-Sky-LLC

CacheTank MCP Server

Official
by July-Blue-Sky-LLC
OverviewSchemaRelated ServersScoreDiscussions
JavaScript
Hybrid

CacheTank MCP Server

Stop re-explaining yourself to AI.

CacheTank is your AI memory layer. Save your identity, projects, decisions, and knowledge once — every AI tool gets it automatically.

Every time you open ChatGPT, Claude, Cursor, Copilot, or Gemini you start from zero. You re-explain who you are, what you are working on, what you have already decided. CacheTank fixes this. Save context once, and every AI-powered tool that supports MCP loads it before your first message.

What CacheTank Does

CacheTank solves the number one frustration of working with AI: repeating yourself. Whether you use one AI tool or ten, CacheTank gives each one your full context without you typing a word.

  • Your identity and role — who you are, what you do, how you think

  • Your active projects — goals, status, decisions, open questions

  • Your accumulated knowledge — things you have learned, patterns you have spotted

  • Your outputs — work you have produced that future conversations should reference

  • Works with every AI — Claude, ChatGPT, Gemini, Cursor, Copilot, Perplexity, and any MCP-compatible tool

Related MCP server: engram

How It Works

  1. Install the CacheTank browser extension (Chrome) or connect via MCP

  2. Save important context as you work — decisions, research, project details

  3. Every AI conversation starts with your full context already loaded

  4. The wisdom cycle promotes repeated concepts automatically

No copy-pasting. No system prompts. No re-explaining. Your context follows you everywhere.

Installation

Claude Desktop (Local)

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "cachetank": {
      "command": "npx",
      "args": ["-y", "cachetank-mcp"],
      "env": {
        "CACHETANK_READ_TOKEN": "your-read-token",
        "CACHETANK_WRITE_TOKEN": "your-write-token"
      }
    }
  }
}

Claude Code (CLI)

claude mcp add cachetank -- npx -y cachetank-mcp

Then set your tokens:

export CACHETANK_READ_TOKEN=your-read-token
export CACHETANK_WRITE_TOKEN=your-write-token

Cursor

Add to Cursor settings (Settings > MCP Servers):

{
  "cachetank": {
    "command": "npx",
    "args": ["-y", "cachetank-mcp"],
    "env": {
      "CACHETANK_READ_TOKEN": "your-read-token",
      "CACHETANK_WRITE_TOKEN": "your-write-token"
    }
  }
}

Windsurf / Cline / Any MCP Client

Use the same npx command:

npx -y cachetank-mcp

Required environment variables:

  • CACHETANK_READ_TOKEN — Your CacheTank read token (get it from the extension or cachetank.com)

  • CACHETANK_WRITE_TOKEN — Optional. Enables saving back to your tank.

Remote Mode (Claude Connectors)

CacheTank also runs as a remote MCP server for browser-based AI tools:

https://cachetank-mcp-77926794635.us-central1.run.app/mcp

MCP Tools

fill_tank

Fetch your personal context for a specific project. Returns your identity, project knowledge, and recent outputs formatted as markdown.

fill_tank({ project: "My Startup" })

cache_it

Save a piece of knowledge, decision, or output to your tank. Saved items become part of your context automatically in future conversations.

cache_it({
  title: "Q1 pricing decision",
  markdown: "Decided on \$29/mo for pro tier based on competitor analysis...",
  project: "My Startup",
  layer: "PROJECTS"
})

MCP Resources

Resource URI

Description

cachetank://context

Your full personal context. Auto-loaded at conversation start.

Why CacheTank Exists

Every knowledge worker using AI tools faces the same problem: context loss. You explain your role, your project, your constraints, your preferences — and then the conversation ends. Next conversation, you start over.

This is not just annoying. It is expensive. Studies show knowledge workers spend 23 minutes re-establishing context every time they switch tools or start a new conversation.

CacheTank is the fix. One knowledge base. Every AI tool. Zero re-explaining.

Pricing

  • Free — 30 saves, full read access, MCP server included

  • Pro — Unlimited saves, $2/week. Full wisdom cycle, all integrations.

Links

Frequently Asked Questions

What is CacheTank?

CacheTank is a personal knowledge layer for AI. It stores your identity, projects, decisions, and knowledge in one place, then serves it to any AI tool via the Model Context Protocol (MCP). Think of it as persistent memory that works across ChatGPT, Claude, Gemini, Cursor, Copilot, and every other MCP-compatible AI tool.

How do I stop repeating myself to AI?

Install CacheTank. Save your context once — your role, projects, preferences, and decisions. Every AI conversation automatically loads your context before the first message. No copy-pasting system prompts. No re-explaining.

Does CacheTank work with ChatGPT?

Yes. CacheTank provides a context URL that any AI tool can read, including ChatGPT. For MCP-compatible tools like Claude and Cursor, context loads automatically. For others, paste your context URL.

What is the Model Context Protocol (MCP)?

MCP is an open standard for connecting AI tools to external data sources. CacheTank uses MCP to give Claude, Cursor, and other compatible tools direct access to your personal context without manual copy-pasting.

Is my data private?

Yes. Your context is stored securely and only accessible via your personal tokens. CacheTank never trains on your data. Read tokens are safe to share with AI tools. Write tokens should be kept private.

How is CacheTank different from custom instructions?

Custom instructions are platform-specific and limited in length. CacheTank is cross-platform, unlimited, and automatically organizes your knowledge by project and priority using a wisdom cycle that promotes important concepts over time.

License

MIT — July Blue Sky LLC

Available Tools

2 tools
cache_itA

Save a piece of knowledge, decision, or output to this person's CacheTank. Use when they make an important decision, reach a conclusion, or produce something worth keeping. They may say "cache it," "save this," or "save to my tank." Saved items become part of future context automatically.

ParametersJSON Schema
NameRequiredDescriptionDefault
markdownYesThe content to save (Markdown)
titleYesShort title (e.g., "Q1 pricing decision")
projectNoWhich project to save under (default: "General")
layerNoSELF = identity (highest priority), PROJECTS = active work, SKILLS = how you think, OUTPUTS = finished work, WORKFLOWS = processes, INBOX = archive. Default: OUTPUTS.

TDQS

A4/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description adds value beyond annotations by stating that saved items become part of future context automatically. Annotations indicate writing intent, and the description complements this with a meaningful side effect.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is concise (3 sentences) and front-loaded with the primary action. Every sentence adds value, making it efficient for an agent to parse.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the schema covers parameter details and annotations are present, the description provides sufficient context for a low-complexity tool. It does not cover return values, but no output schema exists.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

All parameters have descriptions in the schema (100% coverage), so the description does not need to add further meaning. It does not elaborate on parameter usage beyond what the schema provides.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the action ('save') and resource ('CacheTank'), and provides examples of trigger phrases. However, it does not explicitly differentiate from the sibling tool 'fill_tank', which may cause confusion.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description provides explicit guidance on when to use ('important decision, conclusion, or something worth keeping') and includes trigger phrases. It lacks exclusion criteria or alternatives to the sibling tool.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

fill_tankA
Read-onlyIdempotent

Fetch this person's full CacheTank context for a specific project. Returns their identity, active project knowledge, and recent outputs. Use when you need to refresh context or switch projects mid-conversation.

ParametersJSON Schema
NameRequiredDescriptionDefault
projectNoProject name to fetch context for (default: "General")

TDQS

A4.3/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint=true, destructiveHint=false, and idempotentHint=true, which the description aligns with by stating 'Fetch.' The description adds value by specifying the returned content (identity, project knowledge, recent outputs), going beyond the annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is two sentences: the first covers action and returns, the second covers usage. Every sentence is necessary and no information is redundant. It is very concise.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a simple tool with one optional parameter, no output schema, and full annotation coverage, the description provides sufficient context: what it fetches, when to use it, and expected returns. No significant gaps are present.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the input schema already explains the 'project' parameter well. The description reinforces its role by mentioning 'specific project' and 'switch projects,' but does not add new semantic detail beyond the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the action ('Fetch'), the resource ('CacheTank context'), and the scope ('for a specific project'). It also lists the returned data (identity, active project knowledge, recent outputs), making the purpose unambiguous and distinct from the sibling tool 'cache_it'.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description explicitly notes when to use the tool: 'when you need to refresh context or switch projects mid-conversation.' While it doesn't mention when not to use it or alternatives, the provided context is clear and actionable.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections. Dates show when Glama detected each change.

  1. 2 tool updatesv2.1.0
    • First observedcache_it
    • First observedfill_tank
Install Server
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

TDQS

A4.1/5.0
Disambiguation5/5

Each tool has a clearly distinct purpose: 'cache_it' saves a piece of knowledge, while 'fill_tank' retrieves full context. There is no overlap or ambiguity between the two operations.

Naming Consistency4/5

Both tools follow a verb_noun pattern with underscores (cache_it, fill_tank), but 'cache_it' includes the pronoun 'it' instead of a noun, creating a slight inconsistency in structure.

Tool Count4/5

With 2 tools, the server is minimal but appropriately scoped for a simple cache/context retrieval purpose. A few more tools (e.g., delete or list) could be useful, but the count is not inappropriate.

Completeness3/5

The tool surface covers save and retrieve operations, but lacks explicit update or delete functionality, which are notable gaps for full lifecycle management of cached items.

Maintenance

ActivityInactive
ResponsivenessSyncing

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Related MCP Connectors

Related MCP Servers

  • engram

    TushaeBXN
    A
    license
    Not graded
    quality
    C
    maintenance
    Provides persistent, local-first AI memory across sessions via MCP tools for storing, searching, and retrieving context from past interactions.
    1
    MIT

  • UseMyContext

    usemycontext
    A
    license
    Not graded
    quality
    B
    maintenance
    The personal context layer for AI - one user owned profile plus files, read by any MCP client so you never re-introduce yourself.
    1
    MIT

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/July-Blue-Sky-LLC/cachetank-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server