LLM & AI Agent Skills

Edit code → commit → run benchmark → measure metric → keep improvement or revert → repeat forever. Works for any optimization target: LLM training loss, test speed, bundle size, build time, Lighthouse scores, and more. Inspired by Karpathy's autoresearch, pi-autoresearch, and litesearch. This plugin provides two skills that work together. Autoresearch is the core engine (works for any metric), and

Use when auditing a codebase for semantic duplication - functions that do the same thing but have different names or implementations. Especially useful for LLM-generated codebases where new functions are often created rather than reusing existing ones.

Strategies for managing LLM context windows including summarization, trimming, routing, and avoiding context rot Use when: context window, token limit, context management, context engineering, long context.

Every product will be AI-powered. The question is whether you'll build it right or ship a demo that falls apart in production. This skill covers LLM integration patterns, RAG architecture, prompt engineering that scales, AI UX that users trust, and cost optimization that doesn't bankrupt you. Use when: keywords, file_patterns, code_patterns.

AI-powered developer workflows with cost optimization and pattern learning. Run code review, debugging, testing, and release workflows from your terminal or Claude Code. Smart tier routing saves 34-86% on LLM costs. -brightgreen) Empathy Framework is evolving to focus exclusively on Anthropic/Claude to unlock features impossible with multi-provider abstraction: - 📦 Prompt Caching: 90% cost reduct

Screenpipe brand style guide. Reference this when designing UI components, writing copy, or making visual decisions.

This project uses the Model Context Protocol (MCP) to let LLMs directly understand and generate Max patches. Video link. Acknowledgement: the patch being explained is downloaded from here. Text comments in the original file are deleted. Check out the full video where you can listen to the synthesised sounds. The LLM agent has access to the official documentation of each object, as well as objects

Give your agents memory that survives the session, so they stop re-deriving what they already learned and stop repeating past mistakes. Hybrid recall (BM25 + dense + RRF), bi-temporal history, and AES-256-GCM at rest, exposed as 55+ MCP tools that work with any host. 73.8% on LongMemEval's official harness (vs Zep 63.8%, Mem0 49.0%). One binary. One file. No Docker. No Postgres. No cloud. Local-fi

MCP Manager for Claude Desktop A simple web GUI to manage Model Context Protocol (MCP) servers for the Claude Desktop app on MacOS easily. Just follow the instructions and paste a few commands to give your Claude app instant superpowers. The Model Context Protocol (MCP) enables Claude to access private data, APIs, and other services to answer questions and perform actions on your behalf. Learn mor

Search and retrieve biomedical data — genes, variants, clinical trials, articles, drugs, diseases, pathways, proteins, adverse events, pharmacogenomics, and phenotype-disease matching. 15 sources including PubMed, ClinicalTrials.gov, ClinVar, OncoKB, Reactome, UniProt, PharmGKB, OpenFDA, Monarch. Use when asked about gene function, variant pathogenicity, trial matching, drug safety, resistance mechanisms, hereditary syndromes, or literature evidence.

You are an expert LangChain agent developer specializing in production-grade AI systems using LangChain 0.1+ and LangGraph.

Persistent memory for Claude Code. Also: loop detection, scope guards, code analysis, and a local LLM for second opinions. Claude Code forgets everything between sessions and after context compaction. Mini Claude provides: - Checkpoints - Save task state, auto-restore after compaction - Mistake memory - Log errors, get warned next time you touch that file - Decision memory - Log WHY you chose some

Expert guide on prompt engineering patterns, best practices, and optimization techniques. Use when user wants to improve prompts, learn prompting strategies, or debug agent behavior.

aka Vibe Debugging This is an MCP Server and VS Code extension which enables claude to interactively debug and evaluate expressions. That means it should also work with other models / clients etc. but I only demonstrate it with Claude Desktop and Continue here. It's language-agnostic, assuming debugger console support and valid launch.json for debugging in VSCode. 1. Download the extension from re

LLM evaluation and testing patterns including prompt testing, hallucination detection, benchmark creation, and quality metrics. Use when testing LLM applications, validating prompt quality, implementing systematic evaluation, or measuring LLM performance.

Lint any URL for LLM readiness. Get a 0-100 score for token efficiency, RAG readiness, agent compatibility, and LLM extraction quality. Context CLI is an LLM Readiness Linter that checks how well a URL is structured for AI consumption. As LLM-powered search engines, RAG pipelines, and AI agents become primary consumers of web content, your pages need to be optimized for token efficiency, structure

You are an AI assistant development expert specializing in creating intelligent conversational interfaces, chatbots, and AI-powered applications. Design comprehensive AI assistant solutions with natur

Production-ready patterns for building LLM applications. Covers RAG pipelines, agent architectures, prompt IDEs, and LLMOps monitoring. Use when designing AI applications, implementing RAG, building agents, or setting up LLM observability.

GenAI Agent Framework, the Pydantic way Documentation: ai.pydantic.dev FastAPI revolutionized web development by offering an innovative and ergonomic design, built on the foundation of Pydantic Validation and modern Python features like type hints. Yet despite virtually every Python agent framework and LLM library using Pydantic Validation, when we began to use LLMs in Pydantic Logfire, we couldn'

Connect Claude Code to your Robinhood account. Analyze your portfolio, get real-time quotes, and execute trades—all through conversation. Then in Claude Code: 1. Type /mcp → select trayd → click Authorize 2. Sign in with Google 3. Say: "Link my Robinhood account" 4. Approve on your phone 5. Start trading! Instead of clicking through dozens of screens: Important to understand: - Your Robinhood emai

An OpenStreetMap MCP server implementation that enhances LLM capabilities with location-based services and geospatial data. - osm-mcp-server: The main server, available for public use. This server provides LLMs with tools to interact with OpenStreetMap data, enabling location-based applications to: - Geocode addresses and place names to coordinates - Reverse geocode coordinates to addresses - Find

Zero-hallucination answers • Gemini Deep Research • 14 Security Layers • Enterprise Compliance What's New 2026 • Deep Research • Document API • Create Notebooks • Security • Install - 🔍 Query your NotebookLM notebooks — source-grounded, zero-hallucination answers - 📚 Create & manage notebooks programmatically — no manual clicking - 🎙️ Generate audio overviews — podcast-style summaries of your d

Required reading before writing any HogQL/SQL or calling execute-sql against PostHog. Use whenever the user wants to search, find, or do complex aggregations PostHog entities (insights, dashboards, cohorts, feature flags, experiments, surveys, hog flows, data warehouse, persons, etc.) and query analytics data (trends, funnels, retention, lifecycle, paths, stickiness, web analytics, error tracking, logs, sessions, LLM traces).