MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Stop your coding agent burning context on file reads. Delegates big reads to a cheap worker model and returns only the answer — 87-99% saved, measured per call. A hook enforces it in Claude Code; the MCP tools work in any client.
MIT -
Save up to 80% tokens when AI reads code — MCP server for token-efficient code navigation, AST-aware structural reading instead of dumping full files into context window
npm MIT -
code-context-enginegithub.com/elara-labs/code-context-engineupdated this year
Save 94% on AI coding tokens. Index your codebase, agents search instead of reading files. Works with Claude Code, Cursor, Codex, Copilot, Gemini CLI. Local MCP server, free, open source.
pypi MIT -
localthink-mcp-oldgithub.com/H3xabah/localthink-mcp-oldupdated this year
Local Ollama-backed MCP server for Claude Code. Query large files without burning context (~30x token savings)
MIT -
tokennukegithub.com/bigjai/tokennukeupdated this year
Intelligent code indexing MCP server. Tree-sitter AST extraction, hybrid search (FTS5 + vector), call graphs, 10 languages, incremental indexing. Save 99% of tokens.
pypi MIT -
Curated, measurement-driven collection of token-saving MCPs for Claude Code, Codex, Cursor & Windsurf. Each MCP shipped here has passed Humanswith.ai's deterministic-compressor benchmark with documented byte-saving + quality-preservation measurements.
MIT ★ 1 -
MCP server for Angular: compiler-backed template answers, component contracts and version-accurate facts that save AI agents tokens
GPL-3.0 -
MCP server for Angular: compiler-backed template answers, component contracts and version-accurate facts that save AI agents tokens
npm GPL-3.0-or-later -
Save 60-80% tokens when AI reads code — MCP server for token-efficient code navigation with AST-aware structural reading
npm ★ 5 -
@cocaxcode/token-optimizer-mcpgithub.com/cocaxcode/token-optimizer-mcpupdated this year
Orchestration + observability + coach layer for Claude Code token optimization. Measures tool usage, enforces budgets, advises on complementary tools (serena, RTK), and proactively surfaces savings tips.
npm MIT -
json-to-toon-mcp-servergithub.com/your-username/json-to-toon-mcp-serverupdated this year
MCP server for converting JSON to TOON (Token-Oriented Object Notation) format - saves 30-60% on AI tokens
npm MIT -
Context-Canarygithub.com/amarisaster/Context-Canaryupdated this year
A token counter overlay for Claude. Warns you before context compaction hits so you can save what matters.
Apache-2.0 ★ 5 -
Token-saving browser automation MCP server: cached actions, then a local LLM, then Claude. Ground-truth benchmark included; 0 Claude tokens across 20 real runs.
-
skimgithub.com/tonyzdev/skimupdated this year
Skim: A MCP server for Claude Code — skim large outputs, return only schema. Save context, save tokens.
★ 6 -
Zero-cost token optimization engine for AI coding assistants (Claude Code, Cursor, Antigravity, Windsurf). Saves 70-95% tokens via AST skeletons, differential cache, and line slicing.
NOASSERTION ★ 6 -
codebase-pilot-cligithub.com/kalpeshgamit/codebase-pilotupdated this year
Claude Code context engine — pack, compress, and optimize any codebase for AI. Save 60-90% tokens.
npm MIT -
Deterministic, local-first context optimization for LLM prompts: clean, dedupe, structure, and measure token and cost savings before calling a large model.
pypi MIT -
promptthrift-mcpgithub.com/woling-dev/promptthrift-mcpupdated this year
Smart token compression MCP server for LLM apps. Gemma 4 local compression, multi-model cost tracking, intelligent model routing. Save 70-90% on API costs.
MIT ★ 1 -
projectmindgithub.com/Nodemint-dev/projectmindupdated this year
Persistent, compact project memory for AI coding agents — MCP server + CLI. One command, works with Claude Code/Cursor/Windsurf/Gemini/Copilot, and shows you the tokens it saved.
MIT ★ 2 -
local-llm-delegation-mcpgithub.com/mrrodriguez/local-llm-delegation-mcpupdated this year
Intelligent task delegation for AI agents: Offload the "easy" work to your local machine and save premium tokens for the hard stuff.
★ 1 -
SaveContextgithub.com/AlexanderBoger/SaveContextupdated this year
Git LFS for your LLM context — 20x Context Savings — an MCP server that keeps big documents out of the model and returns byte-exact quotes for a fraction of the tokens.
Apache-2.0 ★ 5 -
Token-efficient MCP server for tabular data retrieval. Index CSV/Excel files, query rows, aggregate — 99%+ token savings vs raw file reads.
NOASSERTION ★ 83 -
MCP plugin focused on boosting agent context efficiency and saving tokens
MIT ★ 1 -
panopticongithub.com/krampus-nuggets/panopticonupdated this year
Integrate local LLMs into your codebase. Save on tokens/cost + Always up-to-date documentation.
-
Universal Cybersecurity Model Context Protocol (MCP) Server & Token-Saving Agent Skills for Claude Desktop, Cursor, GPT-4o, and Antigravity. Purple Teaming, WAF Triad Audit, Malware PE Triage, USB Forensics & Caveman Token Optimizer.
MIT -
🚀 Supercharge your AI agents! A cross-platform CLI & Model Context Protocol (MCP) server that parses full codebases into token-efficient, LLM-ready context files. Save tokens, reduce hallucinations, and give your AI the perfect context window.
MIT ★ 1 -
MCP server that connects Claude to local Ollama models, delegating simple tasks to save tokens for complex reasoning
★ 16 -
cache-manager-mcpgithub.com/datawithdanny/cache-manager-mcpupdated this year
Save tokens across AI agent chats with prompt TTL tracking, handoff memories, and cost insights — over MCP.
MIT -
Top Token-Saving AI Coding Tools 2026: Local Memory MCP Cuts Costs 90%
★ 1 -
🧠 Save 70–95% tokens in Claude Code, Cursor, Windsurf & Cline. Local Tree-sitter code graph MCP server: surgical edit context instead of full-file reads, refactor safety checks, and a hook that blocks large file reads. Reduce token usage & cost, zero LLM indexing.
MIT ★ 4 -
classmcpgithub.com/TheDecipherist/classmcpupdated this year
MCP server for AI-assisted CSS development. 77% token savings with semantic class patterns. Supports Tailwind, Bootstrap, UnoCSS, Tachyons.
MIT ★ 8 -
logica-contextgithub.com/Rovemark/logica-contextupdated this year
MCP server that protects your AI context window via sandboxed execution and smart indexing. 97% token savings.
MIT -
Local, private AI chat history search & memory MCP server: index Cursor, Claude, ChatGPT and Cline transcripts with BM25 and save tokens.
MIT ★ 3 -
Instant, accurate Unity API lookups instead of expensive source file reads, saving your agent tokens, context, and hallucinations
MIT ★ 66 -
Local-first agent memory MCP - cross-agent brain with token-saving file diff cache. By Synapse Arrows.
npm MIT ★ 16 -
logzipgithub.com/NailShakurov/logzipupdated this year
Compress logs for LLM analysis — Rust-powered, Python API. 40-60% token savings.
MIT ★ 12 -
Infrastructure for trading AI agents and their humans — save tokens while getting the highest-quality information about the current news background.
MIT -
Claude Code launcher with tools, MCP server and skills that save tokens and give consistent results
MIT -
@iflow-mcp/howardpen9-kimi-code-mcpgithub.com/howardpen9/kimi-code-mcpupdated this year
MCP server for Claude Code × Kimi K2.5 (256K context) — delegate bulk codebase analysis to Kimi, save 90% on token costs
npm MIT -
Contextcoregithub.com/lucifer-ux/Contextcoreupdated this year
ContextCore: An MCP server for Claude (or any AI tool) that enables massive token saving through hybrid search (BM25 + Embeddings)
AGPL-3.0 ★ 22 -
MCP server for Claude Code — multi-AI orchestration across local + cloud LLMs, intelligent backend routing, council-driven workflows, and token-saving file operations.
Apache-2.0 ★ 2 -
mcpgithub.com/supermodeltools/mcpupdated this year
Save up to 40% on agent token costs with code graphs: call graphs, dependency graphs, dead code detection, and blast radius analysis.
MIT ★ 17 -
Instant, accurate Unreal Engine API lookups instead of expensive source file reads, saving your agent tokens, context, and hallucinations.
MIT ★ 99 -
PilotBrowseMCPgithub.com/EndymionLee/PilotBrowseMCPupdated this year
A browser runtime that lets AI agents control your real Chrome browser via MCP. Agents can explore websites, generate operation manuals, and reuse them to save tokens. AI操控你的真实浏览器。
MIT ★ 22 -
Natural Context Provider (NCP). Your MCPs, supercharged. Find any tool instantly, load on demand, run on schedule, ready for any client. Smart loading saves tokens and energy.
npm NOASSERTION ★ 102 -
Save 6-8× tokens on AI code reviews. Builds a structural call graph via Tree-sitter + MCP. Works with Claude, Kimi, Gemini, ChatGPT, Cursor, Windsurf — any AI assistant.
MIT ★ 2 -
webpeelgithub.com/webpeel/webpeelupdated this year
The web data platform for AI agents — fetch, search, crawl, extract, monitor, screenshot. 55+ domain extractors, 65-98% token savings, MCP server included. One API call.
npm AGPL-3.0-only -
context-memgithub.com/JubaKitiashvili/context-memupdated this year
Your AI forgets everything between sessions. This fixes that — 98%+ retrieval accuracy, 100% on LongMemEval, 99% token savings. 44 MCP tools. Fully local, zero cost.
MIT ★ 20 -
db-mcpgithub.com/neverinfamous/db-mcpupdated this year
SQLite MCP Server — Secure Database Administration & Observability with Code Mode. True V8 Isolate Sandbox Unifying 170+ Specialized Tools with Highly Optimized Payloads for 70–90% Token Savings. Includes Dynamic Tool Filtering, Dual-Transport HTTP/SSE, O
npm MIT -
claude-to-agygithub.com/rauls-kjarners/claude-to-agyupdated this year
A lightweight MCP bridge that lets Claude Code delegate heavy tasks to the Antigravity CLI - saving context window and tokens for what matters.
MIT ★ 12 -
larkxgithub.com/utoolslib/larkxupdated this year
AI codebase indexer and MCP server for Claude Code, Cursor, and Copilot. Pre-index your project into a compact graph and measure real token savings with `larkx bench`.
npm MIT -
Save on tokens: your expensive model plans and reviews while cheap pi agents write the code. MCP server with git-worktree isolation, file locks, and a shared board. Works with any MCP client.
Apache-2.0 ★ 1 -
MCP server that saves Claude Code tokens by delegating bounded tasks to local or cloud LLMs. Works with LM Studio, Ollama, vLLM, DeepSeek, Groq, Cerebras.
npm Apache-2.0 ★ 122 -
MCP server for Yandex Direct, Metrika, Webmaster, Wordstat, Search API, Audience and Maps: 127 tools by default; pick platforms via YANDEX_TOOLSETS to save tokens. Guarded writes; optional lazy tool loading.
NOASSERTION ★ 1 -
codebase-pilotgithub.com/kalpeshgamit/codebase-pilotupdated this year
AI context engine for Claude Code, Cursor, Windsurf — pack, compress, and optimize any codebase. Save 60-90% tokens. Web dashboard on port 7456.
MIT ★ 6 -
MCP bridge that lets Claude Code delegate heavy tasks to the Antigravity CLI (agy) — purpose-built tools, model routing with fallback, session continuity, and output truncation to save Claude's context and tokens.
MIT ★ 67 -
Local codebase memory for AI assistants. Because re-explaining your project structure to Claude or Antigravity for the fifth time today gets old. Oh, it saves a LOT on token costs, too.
NOASSERTION -
Full Page Shot: Microsoft Edge extension + MCP server for AI agents. Capture full-page screenshots (PNG/JPEG) of any page. Saves agents tokens by offloading screenshot capture. TypeScript, esbuild, Manifest V3. https://albertovasquezr.github.io/full-page-shot/
MIT -
Native Model Context Protocol (MCP) server providing typed state-action schema arbitration, deterministic action dispatch, and token-saving pre-escalation filters.
MIT -
Every API key your AI apps need, in one local place. Save a token once; Claude, Cursor, VS Code and any MCP app fetch it. No cloud, no login.
MIT