Choose a Template

Use this guide before you create a project.

Choose The Runtime Family

Start with When you need
bub A lightweight Bub app, AG-UI gateway, and AgentSeek lifecycle commands.
deepagents Planning, tool use, sub-agent workflows, sandbox-backed coding agents, or DeepAgents examples with local development.
langchain LangChain or LangGraph app patterns, including RAG, markdown chat, or AG-UI integration.

Every template in the locked standalone catalog exposes lifecycle v2 through .agentseek/lifecycle.toml. The runtime choice still matters because the generated app code is Bub, DeepAgents, LangChain, or LangGraph shaped.

Pick By Project Goal

Goal Choose Why
Build the smallest Bub AG-UI app bub/default It starts a Bub gateway and Vite frontend with the least extra runtime surface.
Wrap a minimal DeepAgents runnable for AgentSeek deepagents/default It binds create_deep_agent(...) through agentseek-langchain.
Run a DeepAgents research workflow deepagents/research It includes search, tool streaming, sub-agent progress, and a React frontend.
Run a DeepAgents content workflow deepagents/content-builder It includes brand memory, skills, subagents, image generation, and streamed UI.
Build a sandbox-backed coding agent deepagents/sandbox It uses create_deep_agent(...) with Daytona by default and keeps the charged LangSmith Sandbox as an alternative.
Build a LangChain AG-UI app langchain/default It keeps the LangChain create_agent(...) shape and binds it through AgentSeek.
Start with a pure LangGraph-style chat UI langchain/markdown-messages It uses langgraph dev, @langchain/react, and markdown message rendering.
Build RAG over OceanBase seekdb langchain/agentic-rag It includes an agentic retrieval tool, ingest command, frontend, and OceanBase seekdb setup.
Learn and inspect hybrid retrieval behavior langchain/agentic-rag-hybrid It includes image ingestion, vector/sparse/full-text/metadata modes, a guided starter pack, visual compare UI, and optional Phoenix traces.
Connect to a remote LangGraph service langchain/cli-remote It bridges a remote LangGraph agent through LangGraphClientRunnable.

Choose AgentSeek-Wrapped Or Framework-Native

Use bub/default, deepagents/default, or langchain/default when you want the generated app to run through the AgentSeek/Bub gateway path.

Use deepagents/research, deepagents/content-builder, deepagents/sandbox, langchain/markdown-messages, langchain/agentic-rag, or langchain/agentic-rag-hybrid when you want a framework-native backend such as langgraph dev, still managed by AgentSeek lifecycle commands.

Choose langchain/agentic-rag-hybrid when you need developers to learn and inspect hybrid retrieval behavior, especially for image-backed knowledge bases where visual similarity, captions, labels, and metadata all affect recall.

Use langchain/cli-remote when the generated project should talk to an already running LangGraph service instead of owning the graph process.

Check Full Details

After you choose a starting point, use the template reference for exact paths and supported create forms.