Topics / #llm-tools

LLM tools projects on GitHub

Utilities and integrations that give large language model applications access to external tools.

As of , the most-starred LLM tools projects on GitHub are Agent-Reach (100.4K stars), QwenPaw (35.9K stars) and fastmcp (28.2K stars). GitHub.Locker tracks 18 LLM tools repositories with 196K stars combined. The fastest-growing on October 19 was Agent-Reach, with +721 stars in a day.

18Projects
196KStars in total
+921Stars in a day

Hot right now

Most stars gained in a day · Oct 19

The all-time top

18 projects · by stars

  1. #1 · Panniantong/Agent-Reach

    Agent-Reach

    Agent Reach is a capability layer that gives an AI agent the ability to actually read and search the internet — Twitter/X, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu, RSS, web pages and more — instead of hitting paywalls, 403s or login walls.

    100.4K
  2. #2 · agentscope-ai/QwenPaw

    QwenPaw

    QwenPaw is a personal AI assistant that you deploy yourself (locally or in the cloud), which remembers context across sessions, runs tasks/automations, and reaches you through chat apps like Discord, Telegram, and WeChat. It is extensible with Skills, Plugins, and MCP tools.

    35.9K
  3. #3 · PrefectHQ/fastmcp

    fastmcp

    FastMCP is an open-source Python framework for building MCP (Model Context Protocol) applications — the layer that connects LLMs to tools and data. It covers servers, clients, and interactive apps.

    28.2K
  1. 4

    a-stock-data

    11Ksimonlin1212/a-stock-data · Python

    It is a self-contained Skill file that bundles raw A-share (Chinese stock market) data from more than two dozen sources into a ready-to-use tool set for AI coding assistants, covering K-lines, research reports, fund flow, news, financials, filings, limit-up boards, ETF options and macro data.

  2. 5

    linkedin-skills

    5Ksergebulaev/linkedin-skills · Python

    A bundle of 12 Claude Code and Codex skills for LinkedIn: they draft posts, comments and replies in your voice, audit drafts before publishing, and build a publishing cadence, all from your terminal.

  3. 6

    adhd

    4.4KUditAkhourii/adhd · TypeScript

    ADHD is a skill for coding agents that spawns several isolated, differently-framed reasoning branches in parallel, then runs a separate critic pass to score, prune, and deepen the survivors, instead of walking one shared chain of thought.

  4. 7

    google_workspace_mcp

    3.4Ktaylorwilsdon/google_workspace_mcp · Python

    Google Workspace MCP Server connects AI assistants to twelve Google Workspace services (Gmail, Calendar, Drive, Docs, Sheets, Slides, Forms, Tasks, Contacts, Chat and more) through 120+ tools exposed over the Model Context Protocol, plus a CLI for scripting.

  5. 8

    Agentlas-OS

    1.6Kagentlas-ai/Agentlas-OS · Python

    Agentlas OS keeps a pool of specialist AI agents in a hub and spins up a temporary orchestrator for each task, splitting work and staffing specialists pulled from your machine, your private cloud, and a public Hub. It is local-first and works with any model, running inside hosts like Claude Code, Codex, Cursor and Gemini.

  6. 9

    TensorFold

    1.4Kashhart/TensorFold · Zig

    TensorFold is an LLM inference engine that serves open-weight language models through an OpenAI-compatible API, running them on Apple Silicon via MLX or on NVIDIA GPUs via CUDA.

  7. 10

    weave

    1.3KAtaraxy-Labs/weave · Rust

    Weave is an entity-level git merge driver: it parses code with tree-sitter into functions, classes, and keys, then merges those entities instead of raw lines, to kill the false conflicts Git invents when independent agents edit the same file.

  8. 11

    Memslides

    937huohua325/Memslides · Python

    MemSlides is a hierarchical memory-driven agent framework that treats slide generation as a stateful authoring process. It uses layered memory (user profile, working, tool) to personalize decks and apply scoped local revisions instead of regenerating the whole presentation.

  9. 12

    CNEquity

    754rootSunc/CNEquity · Python

    CNEquity is local data infrastructure for the Chinese A-share market: it organizes stock quotes, futures, fundamentals, corporate events and capital-flow data from multiple sources into a local Parquet data lake with incremental collection, resumable failures, quality audits and unified querying.

  10. 13

    Awesome-DeepSeek-Harness-Plugins

    574Zhiyuan-Fan/Awesome-DeepSeek-Harness-Plugins

    A curated, daily-updated directory of public plugins, extensions, tools, skills, and integrations for DeepSeek Harness (DSH), an open-source DeepSeek agent harness.

  11. 14

    awesome-jev-live

    377wh000wh000/awesome-jev-live · Python

    Awesome Jev Live is an evidence-graded index of projects built around Jev, TypeSafe AI's System One model — 836 entries covering SDKs, MCP tools, agents, apps and open models, published in 20 languages.

  12. 15

    squeez

    218claudioemmanuel/squeez · Rust

    Squeez is a hook-based token compressor for seven AI coding CLI hosts (Claude Code, Copilot CLI, OpenCode, Gemini CLI, Codex CLI, Pi, Hermes) that intercepts tool output and shrinks bash results by up to 95% without losing exact identifiers.

  13. 16

    iso-24495

    199GaZmagik/iso-24495 · TypeScript

    A set of seven Agent Skills (plus an output style) that teach AI coding agents to write in plain language, following principles inspired by the ISO 24495 Plain Language standard series. They cover core writing rules, legal writing, technical writing, organisational gap analysis, document design, plain-language code, and a text-audit tool.

  14. 17

    Benzi

    169oooscoos/Benzi · HTML

    Benzi is a compiler-backed coding agent: it parses your whole codebase with tree-sitter into a queryable index of symbols, calls, and data/control flow, then uses that map to explain, edit, and verify code changes, including a runtime tracer that settles what static analysis alone can't.

  15. 18

    Matryoshka

    150yogthos/Matryoshka · TypeScript

    Matryoshka (RLM) is an MCP server and CLI toolkit that lets LLM agents analyze documents up to 100x larger than their context window, by having the model emit symbolic 'Nucleus' commands that a logic engine called Lattice executes server-side, instead of chunking text into a vector database.

Often tagged together