Skill UI

Maintains the repository's llms.txt guidance file by auditing current documentation, specs, and structure to ensure the navigation summary matches reality and stays compliant with the official specification.

This guide details advanced strategies for optimizing Claude API usage to reduce latency and cost. Techniques covered include prompt caching (significant cost reduction), strategic model selection (Haiku, Sonnet, Opus) based on task complexity, implementing streaming for perceived speed, and managing token limits. Use this guide when building high-performance, cost-efficient LLM applications that interact with large language models.

Implements a robust, multi-tenant authorization layer for Cohere API v2. This solution enforces Role-Based Access Control (RBAC), manages individual API keys per team, and implements per-team usage limits and daily budget tracking. It is essential for large-scale applications requiring strict resource isolation and cost governance across multiple internal services.

A powerful pre-action hook designed to enforce rigorous checks before AI agents modify code, create files, or execute destructive commands. Instead of relying on superficial self-evaluation, GateGuard forces the LLM to gather concrete facts (like dependencies, data schemas, and impacted modules), significantly improving code reliability and architectural depth. It provides multiple gates for edits, writes, and destructive bash commands.

This guide provides a comprehensive deep dive into migrating applications from OpenAI (GPT) to Anthropic (Claude). It details critical API differences, including changes in system prompt handling, function calling formats, and response parsing. It also demonstrates how to build a robust, multi-provider abstraction layer for seamless LLM switching.

Provides best practices for configuring Large Language Model (LLM) access, specifically for Anthropic Claude, across development, staging, and production environments. This guide details setting up separate API keys, selecting appropriate models (e.g., Haiku, Sonnet, Opus), and managing unique rate limits for robust, scalable, and secure LLM integration in enterprise systems.

This guide provides expert strategies for optimizing GPU inference performance on CoreWeave infrastructure. It covers GPU selection based on workload (LLM, image generation, training), advanced techniques like continuous batching (using vLLM), autoscaling setup (HPA), and benchmarking data. Use this to maximize GPU utilization, minimize latency, and optimize large-scale AI model deployment.

This tool enforces four rigorous coding principles, inspired by Andrej Karpathy's insights on LLM pitfalls. It guides developers to surface assumptions, prioritize simplicity, make surgical changes, and define verifiable goals. Use it when reviewing code, checking diffs, or before committing to ensure high code quality and prevent over-engineering.

A deep dive into handling typed content blocks (text, image, tool_use, thinking, document) in LangChain 1.0. This skill standardizes multi-modal message composition, tool-call iteration, and complex content handling across providers like Claude, GPT-4o, and Gemini, solving key API discrepancies.

This guide provides advanced patterns for composing robust, multi-step Large Language Model (LLM) workflows using LangChain 1.0 core runnables. Learn to implement parallel retrievals (RunnableParallel), conditional routing (RunnableBranch), and state management. It also addresses common pitfalls like silent KeyErrors due to mismatched dictionary shapes, providing solutions using debug probes and type annotations for reliable RAG, classification, and extraction pipelines.

This skill provides a comprehensive guide to building robust LangGraph 1.0 StateGraphs in Python. It covers defining typed TypedDict states with reducers, writing deterministic nodes, and defensively wiring conditional edges. It addresses common pitfalls such as silent graph termination, incorrect state updates using reducers, proper management of recursion limits, and handling evolving checkpoint schemas for reliable LLM workflow orchestration.

This skill defines canonical, ordered middleware patterns for LangChain and LangGraph agents. It addresses critical production-level LLM deployment issues, including PII leakage across tenants, token budget overconsumption, prompt injection defense (Guardrails), and cache key corruption. Use this to build secure, scalable, and cost-aware AI applications.