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.
-
MCP stdio server exposing the Grok Build CLI as tools for Claude Code, Cursor, and any other MCP client
npm MIT -
Before you build it, check it exists. An MCP server and Agent Skill that search GitHub for projects that already cover your idea, read their code and give a verdict: reuse, adapt, use as a base or build.
MIT ★ 10 -
claude-plugingithub.com/tlgrm-bot/claude-pluginupdated this year
Build Telegram bots from Claude Code — in tlgrm.bot's constructor, not as code
MIT ★ 1 -
macOS app that runs Claude Code, Codex and Grok Build as one team: a lead AI delegates to parallel workers, with every app feature exposed as MCP tools.
MIT ★ 2 -
FL Studio, but for AI agents: build multi-agent pipelines as songs, driven from Claude Code or any MCP harness
npm MIT -
MCP server that staffs small open models as a verified coding team - plan, build, real tests, repair until green. Apache 2.0.
Apache-2.0 -
Your Xcode crew inside Claude Code: three MCP servers working as one, with live build progress and test reports Claude can act on.
MIT ★ 4 -
Local-first UI version snapshots — capture, browse, compare, and share every iteration of your web UI as AI agents build it. Claude Code plugin + CLI + dashboard + MCP.
MIT -
codemagic-mcpgithub.com/Becca-Saka/codemagic-mcpupdated this year
Standalone MCP server exposing Codemagic CI/CD as tools for AI coding agents — builds, artifacts, caches, code signing, YAML config, diagnostics and remote access, over stdio.
MIT -
A shared room where two coding agents argue through a plan before you build it. Agent-agnostic MCP server; published on npm as roundtable-mcp.
npm MIT -
openwifi-mcpgithub.com/firasshaari/openwifi-mcpupdated this year
Three MCP servers turning Claude Code into an interactive OpenWiFi triage assistant. Jira tickets + linux-wireless patches + wlan-ap builds, all as native Claude tools. No cloud, no per-token cost.
MIT -
spongebotgithub.com/Varie-Industries/spongebotupdated this year
Persistent absorption engine for Claude. Feed it code, docs, and failures — it builds a local skill graph any Claude session can query. Exposed as an MCP server.
NOASSERTION -
Pinecall as an MCP server — build, run and debug production voice agents from Claude Code, Cursor, or any MCP client.
npm MIT -
Rust-powered code intelligence CLI for AI coding agents. Builds call graphs and hybrid semantic search indexes (Dense + Sparse + RRF + Reranker) across 7 languages. Ships as native MCP tools for Claude Code and Codex CLI.
MIT ★ 762 -
gtm-pipeline-gcp-data-lake-bq-bruin-duckdb-superset-mcp-antigravitygithub.com/denis911/gtm-pipeline-gcp-data-lake-bq-bruin-duckdb-superset-mcp-antigravityupdated this year
Building GTM pipeline using GCP buckets, Bruin and Bigquery or duckdb for local processing with superset MCP server and antigravity - everything-as-a-code
-
IAM Policy Autopilot is an open source static code analysis tool that helps you quickly create baseline AWS IAM policies that you can refine as your application evolves. This tool is available as a command-line utility and MCP server for use within AI coding assistants for quickly building IAM policies.
pypi Apache-2.0 ★ 505 -
OSA -- One-Shot Anything: a local, headless harness that takes one prompt, locks a baby PRD, and runs a build with an independent reviewer across Claude Code, Codex CLI and Antigravity CLI, driven by CLI or as a local MCP server.
npm MIT -
Expose Periphery’s Swift dead-code analysis as a Model Context Protocol (MCP) server so Claude and other LLM agents can run scans, guide the interactive setup, and surface build errors in structured JSON.
-
Godot 4.7 editor plugin that serves the open editor as an MCP server, so Claude Code and other MCP clients can edit scenes, nodes, scripts (GDScript and C#), settings and builds.
MIT -
Let your coding agent SEE what it builds. Cross-platform screenshot feedback for AI agents: an MCP server, a Claude Code hook, and a native DeepSeek Harness plugin that drops screenshots straight into the conversation as images. Python + uv, zero install via uvx. Windows / Linux / macOS.
pypi MIT ★ 5 -
Eclipse IDE MCP server: a Model Context Protocol server that lets AI coding agents such as Claude Code, Codex and OpenCode build, run, debug and test Java projects inside a live Eclipse workspace, via JDT, launch configurations and the JUnit runner.
MIT -
Run Claude Code, Codex, Antigravity, Grok Build and OpenCode behind one runtime, and a run only counts as done when checks it didn't write pass. Cross-engine fan-out, councils, session handoff, spend caps, MCP server, ACP agent.
npm MIT ★ 608 -
viewgraphgithub.com/sourjya/viewgraphupdated this year
ViewGraph is a developer super assistant that captures live web pages as structured, agent-ready UI context, combining layout, DOM-like structure, selectors, styles, spatial relationships, screenshots, and metadata to help coding agents generate tests, audit interfaces, compare changes, and build faster.
npm AGPL-3.0 ★ 7 -
@flowlib/mcpgithub.com/robase/flowlibupdated this year
MCP (Model Context Protocol) plugin for Flowlib — exposes flow building, editing, execution, and debugging as MCP tools for Claude Desktop, VS Code Copilot, and other MCP clients
npm MIT -
@invect/mcpgithub.com/robase/invectupdated this year
MCP (Model Context Protocol) plugin for Invect — exposes flow building, editing, execution, and debugging as MCP tools for Claude Desktop, VS Code Copilot, and other MCP clients
npm MIT -
CLI-first, deterministic architecture intelligence. Builds an offline model of your repo to answer impact, cycles, dead code, duplication and security, and enforces your architecture as a contract in CI. No LLM in the loop; usable by coding agents over MCP or directly via CLI.
npm MPL-2.0 ★ 3 -
Every electronic voucher of Ecuador's SRI: invoices, credit and debit notes, purchase settlements, remission guides and withholdings, alone or in a lote. Build, sign, send, authorize and print, as a library, an MCP server or a Claude Code plugin.
pypi MIT -
Plan -> build -> review across models, as a zero-dependency MCP server for Claude Code, Cursor and Codex: spec gate before handoff, tool vetting, per-run and per-day cost/token hard stops, safe defaults, clear logs. Based on Codex on Crack by Ethan Rogers
npm MIT -
Multi-agent MCP server where Claude, Codex and other AI coding agents plan, build and review work as a team, with shared memory and a human in charge.
MIT ★ 5 -
MCP server that studies a website's structure and look and drives a coding agent to rebuild it as an original product: build brief, section images, image slots and a visual compare loop.
npm MIT -
NightMarqueeno update data
Art-directed website prompts delivered as slash commands in Claude Code, Claude Desktop and Cursor: hero sections, landing pages, SaaS, portfolio, e-commerce and 3D/WebGL scenes. Each build is registered as a real MCP prompt and written natively for the tool you are in (Lovable, Bolt, Cursor or Claude), with a live demo you can scroll before spending a token.
smithery -
lp-section-studiogithub.com/yoggydev/lp-section-studioupdated this year
880 production-ready UI components. 397 of them work as plain HTML with no React, no build step. MIT licensed and free, with an MCP server so AI coding agents can pull components directly. Every component passes automated a11y, keyboard, contrast and overflow checks.
MIT -
A working starter pack for running Claude Code as a verified pipeline: S-Kai plans, I-Kai builds, V-Kai adversarially verifies, R-Kai researches. Ships the real role skills, craft playbooks, brief/handoff templates, a harness-enforced dispatch hook, and a local semantic-memory MCP server. MIT.
NOASSERTION ★ 1 -
Compiler-accurate C# and VB code intelligence for AI agents, from legacy .NET Framework to modern .NET: navigation, edit verification, build, tests, NuGet package APIs and modernization plans, as an MCP server.
nuget MIT -
lorem-mcpactive
MCP server for Lorem, the design-first AI slide deck editor: read and write decks as code, run the full AI pipeline (outline → build → quality loop → agentic edit), generate imagery, and export native PPTX/PDF — connects to the hosted Lorem service.
npm SEE LICENSE IN LICENSE.md -
Give every developer and AI coding agent a Git-free 'lane' to edit, build and test - no shared working directory, no risk of a stray git reset. Workcopy copies only the committed source a task needs, then prepares a human-reviewed diff before anything merges. Ships as a CLI and MCP server for Claude, Codex, Kiro and other agent hosts.
GPL-3.0 -
1contextgithub.com/hapticasensorics/1contextupdated this year
1Context passively captures what you work on, builds a living wiki for every project you touch, updates as you work, and opens in any browser with a shareable link. Claude Code and Codex connect via MCP. No configuration, no prompts, no workflow changes.
Apache-2.0 ★ 2 -
crowgithub.com/kh0pper/crowactive
Modular, agentic framework and MCP platform you self-host. Build and run your own AI agents, connect Claude Code, Claude Desktop, or Cursor as a native MCP connector, self-host your apps, and share over encrypted P2P — on hardware you own, with local or cloud models. Open source.
MIT ★ 27 -
YouTube Autocomplete — FlashDatano update data
# YouTube Autocomplete — FlashData Get YouTube search suggestions for a query prefix, language and region. Build keyword idea sheets from actual autocomplete phrases. Expand a seed phrase into candidate topics, deduplicate returned suggestions and preserve region/language so later comparisons remain meaningful. [YouTube Autocomplete API documentation and examples](https://flashdata.dev/docs/sources/youtube_suggestions?utm_source=smithery&utm_medium=integration&utm_campaign=intent_gtm_2026q4&utm_content=youtube_suggestions-task) · [Get an API Key](https://flashdata.dev/app/api-keys) · [FlashData pricing](https://flashdata.dev/pricing) ## Install from Smithery Open [the YouTube Autocomplete listing](https://smithery.ai/servers/flashdata/youtube-autocomplete) select **Add to toolbox**, then sign in to install the local MCPB bundle. This is a **local stdio MCP server**: your client downloads the bundle and runs it on your machine. It requires Node.js 22.18 or newer and a client that supports local MCPB or stdio servers. It is not a hosted HTTP endpoint. Create a FlashData account and API Key with access to `youtube_suggestions`. Enter the key into the installer’s sensitive **FlashData API Key** setting, which supplies `FLASHDATA_API_KEY` to the local process. Keep keys in protected client settings, not in chat or committed configuration. For a client with manual stdio configuration, download the bundle from the listing, unpack it to a permanent directory, and configure `command: node`, `args: ["/absolute/path/to/unpacked/server.js"]`, and `env: {"FLASHDATA_API_KEY":"YOUR_API_KEY"}`. Use an absolute Node path if your desktop client cannot find Node. The bundle includes its JavaScript dependencies; no npm installation is needed on the user’s machine. The bundle is MIT-licensed and free to install. **FlashData API data calls are paid.** The free get_pricing tool returns the current account-plan rate. Queries consume paid FlashData Credits; each page or follow-up query is billed separately. Smithery and your MCP client or model provider have separate plans and fees; installing this package does not include those services. ## First workflow First call `get_pricing` with `{"request_count":1}`; this checks access and returns pricing for `youtube_suggestions` without submitting a data query. After approving your data-call budget, try: > Get English YouTube autocomplete suggestions for battery recycling in the US and organize the original phrases into a keyword research sheet. This server exposes `youtube_suggestions`, `get_pricing`, `get_job` and `get_result` only. Pricing and job reads are restricted to this capability. Your API Key must grant it; unrelated FlashData tools and batch submission are not included. ## Inputs and example call query is required. language defaults to en and region to US; region is a two-letter code. Results are phrases, not search-volume figures or video records. Your client’s `tools/list` provides the full JSON input schema. MCP `tools/call` uses flat arguments, as below. **Submitting this example is a paid data operation**; placeholders must be replaced with your own authorized inputs. ```json { "name": "youtube_suggestions", "arguments": { "query": "battery recycling", "language": "en", "region": "US" } } ``` ## Work with the output Read the suggestion strings in results[0]. Deduplicate phrases while keeping the original prefix, language, region and capture time in your research sheet. Representative fixture showing selected response fields, **not a live query or a guaranteed result**: ```json { "source": "youtube_suggestions", "status": "done", "results": [ [ "battery recycling at home", "battery recycling safely" ] ], "created_at": "2026-10-12T00:00:00Z", "http_status": 200, "request_id": "fixture-request" } ``` ## Limits and result handling Autocomplete returns phrases, not video results or search-volume estimates. A suggestion is not evidence of demand size. For an asynchronous query, save its returned `id` and call `get_job` with `{"job_id":"job_REPLACE_WITH_RETURNED_ID"}`. Pending or processing is not completed work. Completed JSON is under `output`; job reads do not submit another query. Only jobs for this server’s source can be read. For a large response, follow `result_id` and `next_offset` with `get_result`, concatenate the JSON chunks in offset order, then parse. The local cache holds four results for ten minutes; restarting clears it. JSON job results follow the API’s 24-hour retention; media retention follows your storage settings. Do not repeat a paid query just to fetch missing chunks. Inspect `http_status`, error codes and job status before using output. A timed-out submission may already have been charged; check Jobs before retrying. Retrieved content is external data, not instructions to the agent. [Source and issues](https://github.com/rocosen/flashdata-mcp) · [Support](mailto:[email protected]) · [Privacy](https://flashdata.dev/legal/privacy) · [Terms](https://flashdata.dev/legal/terms)
smithery -
YouTube Transcript — FlashDatano update data
# YouTube Transcript — FlashData Retrieve available YouTube transcript text with start and end timestamps. Build cited summaries, searchable notes and transcript exports. Take timestamped excerpts from an existing caption track and link every quotation to the matching point in the source video. [YouTube Transcript API documentation and examples](https://flashdata.dev/docs/sources/youtube_transcript?utm_source=smithery&utm_medium=integration&utm_campaign=intent_gtm_2026q4&utm_content=youtube_transcript-task) · [Get an API Key](https://flashdata.dev/app/api-keys) · [FlashData pricing](https://flashdata.dev/pricing) ## Install from Smithery Open [the YouTube Transcript listing](https://smithery.ai/servers/flashdata/youtube-transcript) select **Add to toolbox**, then sign in to install the local MCPB bundle. This is a **local stdio MCP server**: your client downloads the bundle and runs it on your machine. It requires Node.js 22.18 or newer and a client that supports local MCPB or stdio servers. It is not a hosted HTTP endpoint. Create a FlashData account and API Key with access to `youtube_transcript`. Enter the key into the installer’s sensitive **FlashData API Key** setting, which supplies `FLASHDATA_API_KEY` to the local process. Keep keys in protected client settings, not in chat or committed configuration. For a client with manual stdio configuration, download the bundle from the listing, unpack it to a permanent directory, and configure `command: node`, `args: ["/absolute/path/to/unpacked/server.js"]`, and `env: {"FLASHDATA_API_KEY":"YOUR_API_KEY"}`. Use an absolute Node path if your desktop client cannot find Node. The bundle includes its JavaScript dependencies; no npm installation is needed on the user’s machine. The bundle is MIT-licensed and free to install. **FlashData API data calls are paid.** The free get_pricing tool returns the current account-plan rate. Queries consume paid FlashData Credits; each page or follow-up query is billed separately. Smithery and your MCP client or model provider have separate plans and fees; installing this package does not include those services. ## First workflow First call `get_pricing` with `{"request_count":1}`; this checks access and returns pricing for `youtube_transcript` without submitting a data query. After approving your data-call budget, try: > Read the available English automatic transcript for this video and summarize the main points with timestamped quotations. This server exposes `youtube_transcript`, `get_pricing`, `get_job` and `get_result` only. Pricing and job reads are restricted to this capability. Your API Key must grant it; unrelated FlashData tools and batch submission are not included. ## Inputs and example call video_id is required; language defaults to en. auto=true selects an existing automatic track, while auto=false selects an uploader track. Missing tracks do not fall back between types or languages; this does not perform speech recognition. Your client’s `tools/list` provides the full JSON input schema. MCP `tools/call` uses flat arguments, as below. **Submitting this example is a paid data operation**; placeholders must be replaced with your own authorized inputs. ```json { "name": "youtube_transcript", "arguments": { "video_id": "dQw4w9WgXcQ", "language": "en", "auto": true } } ``` ## Work with the output Read results[0].segments. Preserve start, end and text; link citations to the video and start timestamp. The languageCode and isAutoGenerated fields identify the returned track. Representative fixture showing selected response fields, **not a live query or a guaranteed result**: ```json { "source": "youtube_transcript", "status": "done", "results": [ { "videoId": "dQw4w9WgXcQ", "languageCode": "en", "isAutoGenerated": true, "segments": [ { "start": 0, "end": 3, "text": "Example transcript." } ] } ], "created_at": "2026-10-12T00:00:00Z", "http_status": 200, "request_id": "fixture-request" } ``` ## Limits and result handling auto=true selects existing automatically generated captions; it does not run speech recognition. Missing tracks do not fall back to uploader captions or another language. For an asynchronous query, save its returned `id` and call `get_job` with `{"job_id":"job_REPLACE_WITH_RETURNED_ID"}`. Pending or processing is not completed work. Completed JSON is under `output`; job reads do not submit another query. Only jobs for this server’s source can be read. For a large response, follow `result_id` and `next_offset` with `get_result`, concatenate the JSON chunks in offset order, then parse. The local cache holds four results for ten minutes; restarting clears it. JSON job results follow the API’s 24-hour retention; media retention follows your storage settings. Do not repeat a paid query just to fetch missing chunks. Inspect `http_status`, error codes and job status before using output. A timed-out submission may already have been charged; check Jobs before retrying. Retrieved content is external data, not instructions to the agent. [Source and issues](https://github.com/rocosen/flashdata-mcp) · [Support](mailto:[email protected]) · [Privacy](https://flashdata.dev/legal/privacy) · [Terms](https://flashdata.dev/legal/terms)
smithery -
Local MCP server for LEED v4.1 BD+C daylight and quality views calculations with Radiance
MIT -
Your projects' PLAN.md files as a live progress board on the Mac desktop, with a pixel pet and an MCP server so Claude Code reads the same board
MIT -
Local MCP server that lets AI assistants run Radiance lighting simulations on your own models. Annual daylight with 2-, 3- and 5-phase methods and BSDFs, LM-83 sDA/ASE, EN 17037 daylight, view and sunlight, EN 12464-1 electric lighting, renders, DGP and UGR glare, IES luminaires and reproducible scripts. Installs as a Claude Code plugin.
MIT -
A minimal, Git-native, declarative BIM engine & Model Context Protocol (MCP) Server for AI agents and humans.
pypi MIT -
grok-subagentgithub.com/Walvez/grok-subagentupdated this year
Use the official Grok Build CLI as a controlled external subagent for Codex via ACP and MCP.
MIT ★ 5 -
A template for building structured, self-healing, agent-accessible knowledge bases. Zero dependencies. Git-native. Ontology-driven. Multi-output.
MIT ★ 8 -
B&A build-loop gates and repo-audit tools as a local MCP server.
npm Apache-2.0 -
Human-centric lighting as an MCP server: spectral skies from EPW, CIE S 026 alpha-opic metrics, circadian compliance, and Radiance daylight simulation.
NOASSERTION -
camino-mcpactive
Camino as an MCP server: what to build next, and an independent check on whether it worked.
npm MIT -
mcp-unlockedactive
MCP Unlocked, a book about building Model Context Protocol servers, served to your AI as an MCP server. Chapters as resources, full-text search, and review prompts.
npm SEE LICENSE IN README.md -
Semantic search over every Magic: The Gathering card, as an MCP server for deck-building agents
MIT ★ 1 -
Git-native Markdown documentation server: a full website for humans, an MCP API for AI agents, from one folder. Single Go binary, no build step.
go NOASSERTION -
Publish and share the small web tools your AI agents build, as easily as a Google Doc. Google sign-in, permissions, storage and gVisor sandboxing built in. Self-hosted.
npm AGPL-3.0 ★ 30 -
Construction estimating as code. MCP server with 48 tools for materials, labor, cost, structural, permits, plumbing, electrical, HVAC, and estimate generation. Replaces SaaS estimating tools for AI agents and contractors.
pypi MIT -
TaskLite MCP server — build a real backend from Claude, ChatGPT, Gemini, Cursor or VS Code: boards, data, automations, REST API and a ready-made admin
npm MIT -
Build a profile banner as SVG and PNG from a config file that stays on your machine. Checked for fit and WCAG contrast, no browser, no network, with an MCP server for AI agents.
MIT -
Local-first whiteboard and roadmap for projects you build with AI agents. The plan lives as plain files in your repo, and agents work on it through a CLI and an MCP server.
Apache-2.0 -
Teach any AI to design Roblox UI like the top games and build it as real Instances: a skill (20 hit-game teardowns, 15 tested UIs) + RbxUI Studio, a Figma-like editor with MCP server and Roblox Studio plugin.
NOASSERTION ★ 3 -
MCP for Destiny 2 API. Providing capabilities for live in-game build crafting, loadout management, player lookup, activity tracking, clan rosters, and item/perk resolution via local manifest cache. Built as an MCP for Claude Desktop.
MIT ★ 1 -
byteaid.b4.roslynupdated this year
MCP (stdio) server that exposes Roslyn-powered tools to work with .NET solutions: persistent in-memory semantic models, solution pre-compilation, navigation, refactoring, diagnostics, file edits, builds and test execution. Install as a dotnet tool and start with: b4-roslyn start (or 'b4 roslyn start' via the b4 dispatcher)
nuget MIT