172 boosters for "lm" — open source, verified from GitHub, ready to install
An AI-powered GitHub issue orchestrator that automatically fetches issues, spawns sub-agents to implement fixes, opens PRs, and monitors review feedback. Ideal for developers wanting to automate bug triage and fix deployment workflows.
This skill guides developers in building LLM-powered applications using Claude and Anthropic SDKs, with smart triggering based on API imports and user intent. Ideal for developers looking to integrate Claude into their projects with language-specific best practices.
mcp-builder is a guide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to integrate with external services and APIs. It helps developers build well-designed tool interfaces in Python or TypeScript for AI-powered applications.
This skill helps you build LLM-powered applications with Claude. Choose the right surface based on your needs, detect the project language, then read the relevant language-specific documentation. Scan the target file (or, if no target file, the prompt and project) for non-Anthropic provider markers
Reality Checker is a skeptical integration agent that enforces evidence-based certification and defaults to 'NEEDS WORK' until overwhelming proof of production readiness is provided. Teams shipping to production benefit from its principled gate-keeping against premature deployments.
A specialized AI agent that automatically detects, classifies, and fixes data anomalies in production pipelines using local SLMs and semantic clustering, with zero data loss guarantee. Data engineers and platform teams benefit most when dealing with broken pipelines that can't afford downtime.
Promptfoo is an LLM evaluation and testing toolkit that helps developers systematically test, benchmark, and validate prompt performance across different models and scenarios. It's essential for teams building LLM applications who need rigorous quality assurance and prompt optimization.
You are a Dropbox specialist for the user's connected Dropbox account. Surface the tool's , , , and the you passed inside when the tool returned them. Never invent a field the tool did not return. Return only one JSON object (no markdown or prose outside it):
Train language models using TRL (Transformer Reinforcement Learning) on fully managed Hugging Face infrastructure. No local GPU setup required—models train on cloud GPUs and results are automatically saved to the Hugging Face Hub. Use this skill when users want to: Use Unsloth () instead of standard
Finds the best models for a task by querying official HF benchmark leaderboards, enriching results with model size data, filtering for what fits on the user's device, and returning a comparison table with benchmark scores.
This skill is for running evaluations against models on the Hugging Face Hub on local hardware. It does not cover: If the user wants to run the same eval remotely on Hugging Face Jobs, hand off to the skill and pass it one of the local scripts in this skill.
Fast, calibration-free weight quantization supporting 8/4/3/2/1-bit precision with multiple optimized backends. HQQ uses to define quantization parameters: The core quantized layer that replaces :
"name": "understand-anything", "description": "AI-powered codebase understanding — analyze, visualize, and explain any project", "homepage": "https://github.com/Lum1104/Understand-Anything",
This skill automates the process of adding, extracting, and managing evaluation results in Hugging Face model cards, supporting multiple data sources including Artificial Analysis API and custom evaluations with vLLM/lighteval. It's valuable for ML practitioners and model maintainers who need to track and display model performance metrics.
1. Fetch homepage HTML (curl or WebFetch) 2. Detect business type (SaaS, Local, E-commerce, Publisher, Agency, Other) 3. Extract key pages from sitemap.xml or internal links (up to 50 pages)
"name": "claude-obsidian", "description": "Claude + Obsidian knowledge companion. Sets up a persistent, compounding wiki vault (Karpathy's LLM Wiki pattern). v1.7 \"Compound Vault\" + v1.8 methodology modes close 5 of 5 priority gaps from the May 2026 compass artifact. Ships: substrate alignment wit
"name": "@mixedbread/mgrep", "version": "0.1.13", "author": "Mixedbread <support@mixedbread.com>",
以下是你所需要生成测试用例的对象的描述，也即来自远程MCP服务器的工具描述。你可以使用调用以下工具。 请首先尽可能全面覆盖并输出所有当前威胁的测试维度，而后为测试目标的每个维度设计测试，对于每个维度至少生成3个测试用例。
Yielding Bear provides a single unified API that routes every LLM request to the cheapest capable model across 16+ providers — saving 60-80% vs calling OpenAI, Anthropic, or Google directly. 1. Get an API key at https://yieldingbear.com/api 2. Set environment variable:
Openmemory JS is a local persistent memory store for LLM applications that enables long-term context retention across Claude Desktop, GitHub Copilot, and other AI platforms. Developers building AI agents and applications benefit from enhanced memory management without external dependencies.
This wiki is maintained entirely by your coding agent. No API key or Python scripts needed — just open this repo in Codex, OpenCode, or any agent that reads this file, and talk to it. Describe what you want in plain English: Or use shorthand triggers:
"name": "mcp-markdownify-server", "description": "MCP Markdownify Server - Model Context Protocol Server for Converting Almost Anything to Markdown", "author": "@zcaceres (@zachcaceres | zach.dev)",
Try these methods in order. Use the first one available: If is wired as a SessionStart hook in , is injected into context automatically at session start. Skip step 1 below. To wire it: or run .
"name": "pro-workflow", "description": "Complete AI coding workflow system. Context engineering, agent teams, 18 hook events, 6 agents, 14 skills, 9 guides, cross-agent support, and searchable learnings.", "name": "Rohit Ghumare",
