Tokensave

v4.0.2Coding Agentsstable

The most comprehensive code intelligence MCP server for AI coding agents. 40+ tools, 30+ languages, 9 agent integrations. Pre-indexed semantic knowledge graphs for instant code understanding — fewer tokens, fewer tool calls, 100% local.

tokensavemcpai-integration
Share:
133
Stars
0
Downloads
0
Weekly
0/5

What is Tokensave?

Tokensave is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to most comprehensive code intelligence mcp server for ai coding agents. 40+ tools, 30+ languages, 9 agent integrations. pre-indexed semantic knowledge graphs for instant code understanding — fewer token...

The most comprehensive code intelligence MCP server for AI coding agents. 40+ tools, 30+ languages, 9 agent integrations. Pre-indexed semantic knowledge graphs for instant code understanding — fewer tokens, fewer tool calls, 100% local.

This server falls under the Coding Agents and Developer Tools categories on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • The most comprehensive code intelligence MCP server for AI c

Use Cases

Reduce token usage in AI coding with pre-indexed semantic knowledge graphs.
Support 30+ programming languages with 40+ intelligent tools.
LicenseMIT License
Languagerust
Versionv4.0.2
UpdatedMay 21, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

Manual Installation

npx tokensave

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Tokensave

Tokensave is a Rust-based MCP server that builds a pre-indexed semantic knowledge graph of your codebase using a local libSQL database, so AI coding agents can query symbols, call graphs, and code relationships instead of repeatedly reading files with grep and glob. It exposes 80+ MCP tools covering semantic search, call graph traversal, dead code detection, atomic edit primitives, and code health metrics across 50+ programming languages including Rust, Go, Java, Python, TypeScript, C/C++, Swift, and shader languages. With support for 12+ agent integrations — Claude Code, Codex, Gemini CLI, Cursor, Copilot, Cline, and more — it is designed to dramatically reduce token consumption on complex coding tasks while running entirely locally with no external API calls.

Prerequisites

  • Rust 1.70+ if installing via cargo (or use Homebrew/Scoop/prebuilt binary)
  • macOS, Linux (x86_64 or ARM64), or Windows
  • At least one supported AI coding agent: Claude Code, Cursor, Gemini CLI, Codex CLI, GitHub Copilot, or similar
  • A codebase to index (any language supported in the lite/medium/full tier)
1

Install tokensave for your platform

Choose the installation method for your OS. Homebrew is recommended on macOS, Scoop on Windows, and cargo on any platform. Prebuilt binaries are also available on the GitHub releases page.

# macOS (Homebrew)
brew install aovestdipaperino/tap/tokensave

# Windows (Scoop)
scoop bucket add tokensave https://github.com/aovestdipaperino/scoop-bucket
scoop install tokensave

# Any platform (cargo - full tier with 50+ languages)
cargo install tokensave

# Minimal binary (lite tier)
cargo install tokensave --no-default-features
2

Register tokensave with your AI coding agent

Run tokensave install to auto-detect installed agents and register the MCP server. Pass --agent to target a specific agent. For Claude Code, this also installs PreToolUse hooks that block wasteful Explore sub-agents.

# Auto-detect all installed agents
tokensave install

# Target a specific agent
tokensave install --agent claude    # Claude Code
tokensave install --agent cursor    # Cursor
tokensave install --agent gemini    # Gemini CLI
tokensave install --agent codex     # OpenAI Codex CLI
tokensave install --agent copilot   # GitHub Copilot
3

Index your project

Navigate to your project root and run tokensave init. This builds the local libSQL knowledge graph from your source files. The index stays fresh automatically via a post-commit hook and on-demand sync.

cd /path/to/your/project
tokensave init
4

Verify the index and check token savings

Use the CLI to confirm the index is up to date and view a savings dashboard showing how many tokens and tool calls tokensave has saved versus raw file scanning.

# Check index status
tokensave status

# View token savings dashboard
tokensave dashboard

# View cumulative savings across sessions
tokensave savings --all
5

Search your codebase from the CLI

Use tokensave search to run a semantic query directly from the terminal, validating that the index works before involving an AI agent.

tokensave search "authentication flow"

Tokensave Examples

Client configuration

After running tokensave install --agent claude, the MCP config is written automatically. The equivalent manual config for Claude Code's .mcp.json looks like this:

{
  "mcpServers": {
    "tokensave": {
      "command": "tokensave",
      "args": ["serve"]
    }
  }
}

Prompts to try

Once tokensave is indexed and registered, AI agents use the MCP tools automatically. You can also ask the agent directly.

- "Find all callers of the authenticate() function in this codebase"
- "Show me the full call graph for the payment processing module"
- "What functions are dead code in the src/utils directory?"
- "Search for all code related to rate limiting and return the relevant symbols"
- "What is the code health score for this project and where are the riskiest files?"

Troubleshooting Tokensave

tokensave init completes but the agent still reads files with grep/glob

Ensure tokensave install was run for your specific agent after init. For Claude Code, verify the PreToolUse hook is installed by checking CLAUDE.md in the project root for tokensave instructions.

cargo install fails with compilation errors

tokensave requires Rust 1.70+. Run rustup update stable to update your toolchain. If you want to avoid compiling, use the prebuilt binary from the GitHub releases page or Homebrew.

Index becomes stale after many file changes

Run tokensave init again to trigger a full re-index, or call the reindex MCP tool from your agent. Install the global git post-commit hook with tokensave install --git-hook yes to keep the index automatically fresh after every commit.

Frequently Asked Questions about Tokensave

What is Tokensave?

Tokensave is a Model Context Protocol (MCP) server that most comprehensive code intelligence mcp server for ai coding agents. 40+ tools, 30+ languages, 9 agent integrations. pre-indexed semantic knowledge graphs for instant code understanding — fewer tokens, fewer tool calls, 100% local. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Tokensave?

Follow the installation instructions on the Tokensave GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.

Which AI clients work with Tokensave?

Tokensave works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Tokensave free to use?

Yes, Tokensave is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Coding Agents MCP Servers

Explore all coding agents servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "tokensave": { "command": "npx", "args": ["-y", "tokensave"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Tokensave?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides