UitbreidenOS/UitKit - Página 19

session-memory

Claude Code session checkpoints and state restoration. Serialize working directory, uncommitted git status, recently modified files, and active task description into checkpoint.json to resume context later

Idioma do texto original: inglês

advanced-tool-use

Master advanced tool use patterns including chaining, parallel calls, and error recovery for complex Claude Code workflows

Idioma do texto original: inglês

agent-construction

Multi-agent architecture, orchestrator patterns, tool design, agent loops, memory, error handling, handoffs

Idioma do texto original: inglês

agent-handoff

Implement structured handoffs between agents with JSON-based state transfer, conflict detection, and acknowledgment protocols

Idioma do texto original: inglês

agent-memory

Agent memory and context management: design persistent memory systems, manage context windows, implement summarization strategies, and build stateful agents that learn across sessions

Idioma do texto original: inglês

agent-sdk

Build AI agents using the Claude Agent SDK with tool use, memory, and multi-step orchestration

Idioma do texto original: inglês

agent-supervisor

Build orchestrator agents that manage multi-agent workflows, enforce SLAs, handle escalations, and maintain overall task completion

Idioma do texto original: inglês

agent-tracing

Build observability into multi-agent systems with distributed traces, trace correlation, breakpoints, and execution replay

Idioma do texto original: inglês

artifact-first-workflow

Claude Code Artifact-First Workflow: single source of truth specifications. Draft, version, and edit specifications (SPEC.md) first, treating the actual codebase implementation as a compilable/generated output

Idioma do texto original: inglês

caveman-token-reduction

Claude Code Caveman mode: token-efficient communication, compressed responses, minimal output for context-heavy sessions, terse acknowledgments, and context budget optimization

Idioma do texto original: inglês

claude-api

Anthropic Claude API: prompt caching, streaming, tool use, batch processing, model selection, cost optimization

Idioma do texto original: inglês

claude-batch

Run large-scale batch processing with Claude API for bulk text analysis, classification, and generation

Idioma do texto original: inglês

crewai

Build multi-agent crews with CrewAI framework for collaborative AI task execution

Idioma do texto original: inglês

dag-scheduler

Build, validate, and execute directed acyclic graph (DAG) topologies for multi-agent workflows with automatic parallelization and cycle detection

Idioma do texto original: inglês

dispatch-background-jobs

Claude Code Dispatch: run long-running tasks as background jobs with progress polling, parallel sub-agent fan-out, result aggregation, and automatic retry on failure

Idioma do texto original: inglês

enterprise-search

Enterprise search and knowledge retrieval: design multi-source search systems, implement hybrid search (semantic + keyword + structured), build knowledge graphs, handle access control, and create unified search across docs, wikis, code, and databases

Idioma do texto original: inglês

graph-augmented-context

Claude Code Graph-Augmented Context (GAC): local SQLite codebase knowledge graph builder. Maps files, classes, functions, and import/dependency edges for advanced codebase RAG and contextual navigation

Idioma do texto original: inglês

langgraph

LangGraph stateful agent framework: StateGraph, nodes, edges, conditional branching, persistence/checkpointing, human-in-the-loop, multi-agent subgraphs

Idioma do texto original: inglês

llm-benchmark

LLM benchmarking: design benchmark suites, establish baselines, run A/B model comparisons, detect regressions, track production quality metrics, and create leaderboards for model selection

Idioma do texto original: inglês

llm-eval

LLM evaluation: build evaluation datasets, choose metrics (RAGAS, G-Eval, LLM-as-judge), run automated evals, monitor production quality, and detect regressions

Idioma do texto original: inglês

managed-agents

Deploy and manage Claude agents with lifecycle management, monitoring, and scaling

Idioma do texto original: inglês

mastra

Mastra AI: TypeScript agent framework — agents, workflows, tools, memory, RAG, evals — for building production-grade AI applications

Idioma do texto original: inglês

mcp-dynamic-discovery

Claude Code MCP Dynamic Discovery: zero-config external tool integration. Automatically scans system configuration directories for active Model Context Protocol (MCP) servers and configures them locally

Idioma do texto original: inglês

mcp-server-builder

Build custom MCP servers: expose tools, resources, and prompts via stdio or HTTP/SSE transport — connect proprietary data to Claude Code

Idioma do texto original: inglês

modal

Build serverless AI pipelines with Modal for cloud-based compute and inference

Idioma do texto original: inglês

multi-agent-memory

Build shared, versioned memory systems for multi-agent teams using blackboard pattern, conflict resolution, and eventual consistency

Idioma do texto original: inglês

prompt-caching

Optimize Claude API costs and latency with prompt caching strategies

Idioma do texto original: inglês

prompt-engineering

Prompt engineering for production: few-shot prompting, chain-of-thought, system prompt design, output formatting, temperature settings, and evaluation — for Claude and other LLMs

Idioma do texto original: inglês

ptc

Implement prompt-to-code workflows for rapid prototyping and code generation

Idioma do texto original: inglês

rag-architect

RAG system design: chunking strategies, embedding model selection, vector store choice, retrieval patterns, reranking, evaluation — production-grade retrieval-augmented generation

Idioma do texto original: inglês

remote-control-headless

Claude Code Remote Control: manage headless Claude Code sessions from anywhere, CI/CD integration, remote orchestration from mobile/laptop, session attach/detach, and remote execution patterns

Idioma do texto original: inglês

session-handoff

Claude Code Session Handoff: structured context export/import between sessions, state serialization, continuation prompts, cross-session knowledge transfer, and resumable workflows

Idioma do texto original: inglês

skill-composition

Skill composition and chaining: design multi-skill workflows, chain skill outputs as inputs, build reusable skill pipelines, and orchestrate complex tasks from atomic skill building blocks

Idioma do texto original: inglês

swarm-sandbox

Safe isolated testing environment for multi-agent swarm topologies before production deployment

Idioma do texto original: inglês

vercel-ai-sdk

Vercel AI SDK: streaming AI responses to Next.js, useChat/useCompletion hooks, tool calls, generative UI, multi-provider routing (Claude, OpenAI, Gemini)

Idioma do texto original: inglês

browser

Browser automation with Playwright: web scraping, UI testing, form filling, screenshot capture, PDF generation, session recording

Idioma do texto original: inglês

claude-design-branding

Create consistent brand systems with Claude Design including logos, colors, and typography

Idioma do texto original: inglês

claude-design

Build production-ready UIs and prototypes with Claude Design

Idioma do texto original: inglês

claude-design-tokens

Manage design tokens for consistent theming across Claude Design projects

Idioma do texto original: inglês

claude-design-workflow

Build end-to-end design workflows with Claude Design for rapid UI/UX iteration

Idioma do texto original: inglês