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.
-
@token-budget/mcp-servergithub.com/tachika710/token-budgetupdated this year
MCP 服务器 — 把 token-budget 核心能力暴露给所有支持 MCP 的客户端(Claude Desktop / Cursor / Trae / Cline / DSH 等)
npm MIT -
🔌 Exposes agent token budget over MCP — Part of the MEOK A2A Substrate
pypi MIT -
agent-token-budgetactive
Zero-dependency context window compressor, code signature folder, JSON minifier & token estimator for AI coding agents and MCP.
npm MIT -
MCP server exposing token-budget as callable tools (create a budget, add messages, get the strategized context, explain a decision, truncate a tool result) — for testing and driving the library interactively from Claude Code, Claude Desktop, or any other
npm MIT -
budget-aware-mcpgithub.com/doorman11991/budget-aware-mcpupdated this year
Model-agnostic code memory MCP server. Budget-aware graph retrieval for AI agents — sub-millisecond queries, token budgeting, no embeddings, no API keys. Built on CodeGraphContext for 155-language tree-sitter indexing.
npm MIT -
@barbozaa/memory-tokengithub.com/barbozaa/memory-tokenupdated this year
MCP server + Cursor hooks for governed workspace memory: token-budget context packs, propose/confirm flow, typed link graph, and local RAG with embeddings.
npm MIT -
@webfacemedia/budget-planner-mcpgithub.com/webface/budget-plannerupdated this year
MCP server that exposes the Budget Planner app's transaction-data tools to Claude Desktop, LM Studio, and other MCP clients. Bearer-token authenticated; tokens minted in /settings.
npm MIT -
Cuts Claude Code and LLM token usage by 60-85% via AST navigation, deduplication, and context-budget optimization.
-
@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 -
Static worst-case token-budget analysis for LLM-agent workflows + signed budget certificates.
npm -
Validation gates, plan checks, and token-budget guardrails for AI coding agents.
npm MIT -
Token and context budget for AI agents. Tracks spend, estimates the cost of reads, and tells an agent the cheaper path before it burns its budget.
npm MIT -
@the-kitchen/brain-kitupdated this year
Token-budget aware memory engine + MCP server for AI agents — GBrain alternative with RRF hybrid search
npm MIT -
Token-budgeted codebase context packer for LLMs: XML, Markdown, llms.txt formats. CLI, MCP server, and GitHub Action. Zero runtime dependencies.
Apache-2.0 -
hyperliquid-hip-1-deployment-calculatorgithub.com/0xkaz/hyperliquid-hip-1-deployment-calculatorupdated this year
HIP-1 token deployment validator and budget calculator for Hyperliquid L1. Real-time parameter validation, budget estimation, MCP integration, and E2E tested.
MIT ★ 2 -
Token budget auditor for agent context: per-file and per-directory token estimates for a repo, the top offenders eating the window, and concrete exclusion advice, against a configurable budget.
npm MIT -
muster-mcpgithub.com/JoasASantos/muster-mcpupdated this year
Rust core + MCP server for running many LLM agents in parallel without blowing your token budget — hard budget caps, automatic work dedup, and deterministic result merging.
★ 5 -
contextcache-mcpgithub.com/pandolfipedro/contextcache-mcpupdated this year
Budget-aware MCP server that builds minimal Context Packs for AI coding assistants - dependency graph, token budget, local SQLite, no paid APIs.
MIT ★ 1 -
token-lens-mcpgithub.com/mdfifty50-boop/token-lens-mcpupdated this year
MCP server for context window optimization — audit token budgets, compress schemas, optimize tool loading
npm -
mcp-ynabgithub.com/bold-minds/mcp-ynabupdated this year
Read-only Model Context Protocol server for YNAB budgeting. OS-keyring token storage, host-locked HTTP, rate-limited, Go.
go MIT -
MCP server for @mukundakatta/agentbudget — token + dollar budget caps for AI agents over stdio. Drop into Claude Desktop, Cursor, Cline, Windsurf, Zed.
MIT -
Local, deterministic, token-budgeted code context for coding agents: cited source excerpts and a code graph inside a token budget. CLI, MCP server and VS Code extension.
npm MIT -
Token-budgeted, self-healing browser automation MCP server: returns only what changed, ~99% fewer tokens than full-tree servers.
MIT -
A token-optimization and memory layer between agent and model: dedups tool output, distills old turns, prunes to a token budget. Proxy, MCP server, CLI, or library
Apache-2.0 ★ 1 -
Task-adaptive repository context engine for coding agents: local MCP, exact token budgets, and dependency-aware retrieval.
MIT ★ 1 -
fittok — relevant-code retrieval within a token budget (MCP server + CLI + library)
pypi MIT ★ 3 -
slicegrepgithub.com/haxo98098/slicegrepupdated this year
grep that returns ranked, token-budgeted code slices — built for LLMs and coding agents. CLI + Python library + MCP server, zero dependencies.
MIT ★ 1 -
Claude Code quota planner: a status line for your 5h/7d windows — or your USD budget on pay-per-token auth — plus an MCP server that says whether a task fits before you start it.
MIT -
VS Code extension that shows what is left on your Bifrost virtual key (budget, tokens, requests) in the status bar, warns before it runs out, and gives coding agents the same numbers through an MCP tool. Unofficial. Built for the Global AI Hackathon, PS04.
Apache-2.0 -
A local Model Context Protocol (MCP 2.x) server exposing RepoDigest's AST signatures, call graphs, and token-budgeted context packing to AI coding assistants.
MIT ★ 1 -
Token-slimming AST code outliner, log compressor & context budget analyzer for AI coding agents (Cursor, Claude, Cline, Antigravity). Slashes token usage & LLM API bills up to 85%.
MIT -
contextpulse-mcpgithub.com/DIYA73/contextpulse-mcpupdated this year
Real-time MCP server that tracks context budget, detects agent loops, and alerts before token overflow — for Claude Code, Cursor, and any MCP-compatible AI tool.
★ 1 -
Vision models pay for tiles, not megapixels. Per-provider token math, exact + budget resize, zero-dep PNG/JPEG codecs, Claude Code hook, MCP server.
npm MIT -
16ane sandbox walletgithub.com/bytesbrains/16aneno update data
Budget-capped agent wallet: pay x402 services in tANE, a free testnet token with no cash value
npm -
Symbol Delta Ledger (SDL-MCP) is a policy-centered context budget layer for coding agents: Symbol-graph intelligence combined with precision tools. It turns sprawling codebases into compact, high-signal context that saves tokens, speeds up workflows, and improves agent output.
npm NOASSERTION ★ 500 -
context-squeezegithub.com/yousudip/context-squeezeupdated this year
Deterministic context-compression MCP server for Claude: squeeze codebases, files, and logs to fit token budgets using tree-sitter and tree-walking — no LLM in the loop. (Rust)
MIT ★ 1 -
MCP server and library that help agents view and annotate multimodal data within a token budget
pypi Apache-2.0 -
Filters an agent's memories to fit a token budget. Local, HTTP, MCP, Docker.
MIT -
notebookllmgithub.com/Yasirrazaa/notebookllmupdated this year
Jupyter notebooks, made agent-ready. 8+ formats · MCP server · async execution · token budgeting.
MIT ★ 4 -
aperture-mcpgithub.com/mohamedh1998/apertureupdated this year
Token-budget-aware visual feedback loop for AI coding agents
npm MIT -
ComparEdge LLM Costno update data
Token cost math for LLM API calls: current per-million-token rates for 69 models across 17 providers, verified and date-stamped. Estimates single calls, comparisons across models, and monthly budgets. Zero dependencies, no API key.
smithery -
Pre-Execution Token Budget Firewall & Cost Telemetry Sentinel for AI Coding Agents
npm MIT -
Token-budgeted MCP server for coding agents: tree-sitter maps, call graphs, and project docs over stdio.
MIT ★ 1 -
A ranked, token-budgeted map of any codebase for AI coding agents. CLI, MCP server and Python API.
pypi Apache-2.0 -
Local code-intelligence engine for AI agents — indexes a repository into a PostgreSQL code graph and serves token-budgeted context (symbols, effects, contracts, blast radius, similar code) over MCP and HTTP.
Apache-2.0 ★ 7 -
Track Claude Code, Codex and LLM API token usage from the Windows and Linux system tray: session budgets, real Codex limits, per-day API-equivalent costs and local history. Multi-provider local proxy and read-only MCP server.
MIT -
tokenslimgithub.com/himanshudhawale/tokenslimupdated this year
Shrink the token cost of context you feed to LLMs — count tokens, estimate cost across models, and slim files (with a CI budget guard) before you send them.
MIT -
mcp-change-risk-assessorgithub.com/hardik-369/mcp-change-risk-assessorupdated this year
Ultra-low token MCP server for autonomous code change risk assessment. No API keys, works offline, 50-100 token budget.
npm MIT -
MCP for generating ranked, token-budgeted code structure maps using Tree-sitter AST analysis and PageRank
npm -
Live web context for AI agents: task-focused Markdown, a token budget, and visible omissions. Local MCP. Package identity: ff-occam.
npm AGPL-3.0 -
jellybeangithub.com/rayhankhilji/jellybeanupdated this year
A token-frugal MCP server that gives coding agents structural understanding of a codebase: repo maps, symbol outlines, ranked search, import tracing, and parsed diagnostics — every result under a token budget it honours.
MIT ★ 4 -
agentic-scopegithub.com/jessn-dev/agentic-scopeupdated this year
A directory-as-context standard plus a read-only MCP server that gives AI agents live, structured, token-budgeted awareness of a multi-project workspace.
MIT ★ 2 -
distilno update data
Context optimization middleware for LLM agents — a dynamic tool registry, result masking, token budgeting and smart compaction, so a long session stays inside its window instead of failing for length.
smithery -
Local code-context engine for AI coding agents: structural answers within a token budget.
npm Apache-2.0 ★ 4 -
prismgithub.com/0xCondor/prismupdated this year
Optimal context selection for AI coding agents - multi-signal code graph + token-budget knapsack, served over MCP. 100% local.
MIT -
@tinydarkforge/codicilgithub.com/tinydarkforge/codicilupdated this year
Local-first memory and assertion ledger for AI coding agents. Confidence-weighted, contradiction-aware, token-budgeted context over MCP.
npm MIT -
nesiftgithub.com/scottgl9/nesiftupdated this year
Fast, local semantic search over web content for AI agents. Hybrid BM25 + potion-retrieval-32M embeddings, cross-page dedup, token-budget mode, MCP server, SearXNG bridge. ~90% fewer tokens than raw web_fetch.
GPL-2.0 ★ 4 -
unblinkgithub.com/ChristopherDavenport/unblinkupdated this year
Pure-Go browser for AI: fetch pages, optionally render JS (no Chromium), get clean token-budgeted Markdown over MCP
go MIT ★ 8 -
memory-safegithub.com/eljoesb/memory-safeupdated this year
MCP memory server for Claude Code — knowledge graph in SQLite, local ONNX embeddings, token-budgeted retrieval
npm MIT -
lodestar-contextgithub.com/kostyadudin/lodestarupdated this year
Local-first repository context tool for LLM agents — MCP server with SQLite/FTS5 index, symbol-aware ranking, token-budgeted packs, and durable memory.
pypi MIT