Skip to content
Get packs

Stack for ai builders

The prompt & agent builder stack

For prompt engineers, AI product builders and anyone moving from chatting to building. Start with the best open guides and our prompt tools, prototype visually, graduate to agent frameworks, run models locally and give agents memory and retrieval.

Open source
18
Agent skills
3
Browser tools
3
Total pieces
29
Download stack kit

ZIP: setup README, install helper, the skill folders (with licences) and free prompts.

How it connects

  1. 01

    Learn the patterns

    Work through the Prompt Engineering Guide and cookbooks; practise with our prompt builder and checker.

  2. 02

    Prototype

    Sketch the agent in Langflow or AnythingLLM with your documents.

  3. 03

    Productionise

    Rebuild the flow in LangGraph, CrewAI or the Agents SDK; optimise prompts with DSPy.

  4. 04

    Own your stack

    Serve models with Ollama/llama.cpp, prepare documents with Docling/MarkItDown, add LightRAG and memory.

1 · Master prompting

Fundamentals and tools that make every model work better.

Prompt Engineering GuideOpen sourceThe most complete open guide to prompt and context engineering.★ 79.8k · MITClaude CookbooksOpen sourceOfficial recipes for getting great results from Claude.★ 54.1k · MITOpenAI CookbookOpen sourceExamples and guides for the OpenAI API.★ 76.9k · MITPPrompt Structure BuilderBrowser toolAssemble role, context, task and format into one prompt.PPrompt Quality CheckerBrowser toolScore a prompt and get specific fixes.PPrompt Variable Template BuilderBrowser toolTurn prompts into reusable templates with variables.SStructured JSON outputPromptReliable JSON output from chat models.Free promptGoogle AI StudioAI toolFree playground to compare prompts on Gemini models.FreemiumClaude ConsoleAI toolAnthropic's workbench with a prompt improver and evaluations.Paid

2 · Prototype agents visually

Wire up agents and workflows before writing much code.

LangflowOpen sourceVisual builder for agents and RAG flows.★ 156k · MITAnythingLLMOpen sourcePrivate all-in-one app for chatting with documents and agents.★ 67.8k · MITLibreChatOpen sourceSelf-hosted multi-model chat with agents, MCP and skills.★ 46.6k · MIT

3 · Build with frameworks

Code-first frameworks for production agents.

LangChainOpen sourceThe most widely used agent framework.★ 149k · MITLangGraphOpen sourceDurable, stateful agent graphs.★ 44.3k · MITCrewAIOpen sourceRole-playing multi-agent teams.★ 60.4k · MITOpenAI Agents SDKOpen sourceLightweight multi-agent workflows.★ 30.5k · MITDSPyOpen sourceProgram — don't hand-tune — prompts with optimisers.★ 38.9k · MITSkill CreatorAgent skillAnthropic's skill for creating and testing your own skills.Agent skill · MITBuild MCP Server GuideAgent skillGuides an agent through building an MCP server.Agent skill · Apache-2.0Context EngineeringAgent skillDesigns what context an agent gets, and when.Agent skill · MITBrowser UseOpen sourceAgents that operate real websites.★ 120k · MITpromptfooAI toolTest prompts and agents against cases before shipping.Open sourceLangfuseAI toolOpen-source tracing, prompt management and evals.Open source

4 · Run, remember and retrieve

Local models, memory and retrieval that make agents useful over time.

OllamaOpen sourceRun open models locally with one command.★ 186k · MITllama.cppOpen sourceFast local inference on CPUs and GPUs.★ 133k · MITLightRAGOpen sourceSimple, fast retrieval-augmented generation.★ 40.5k · MITDoclingOpen sourceConvert PDFs and office docs into clean inputs for RAG.★ 70k · MITMarkItDownOpen sourceConvert any file to Markdown for LLMs.★ 193k · MITMemPalaceOpen sourceOpen-source long-term memory for agents.★ 60.2k · MIT

Setup checklist

  • Python 3.10+ and Node.js 20+
  • Install Ollama and pull one small and one large model
  • Get an API key for at least one frontier model for comparison
  • Create an evaluation set of 20 real tasks before optimising prompts