deepagents-typescript-quickstart
langchain-ai/langchain-skills
Scaffold a minimal TypeScript Deep Agent locally using provider-native web search instead of Tavily.
What is deepagents-typescript-quickstart?
Quickly set up a Deep Agent in TypeScript by following LangChain's official quickstart. This skill creates a research agent that uses your chosen provider's built-in web search (Anthropic, OpenAI, or Google) without requiring a separate search API. Use when you want to build or experiment with a Deep Agent locally.
- Scaffold a new Deep Agent project directory with TypeScript setup
- Implement the research-agent shape from LangChain's official quickstart docs
- Replace Tavily with provider-native web search (Anthropic, OpenAI, or Google)
- Prompt you to select a model using `provider:model` format (e.g., `anthropic:claude-sonnet-5`)
- Run a research example and display output
- Point to customization and Managed Deep Agents for next steps
How to install deepagents-typescript-quickstart
npx skills add https://github.com/langchain-ai/langchain-skills --skill deepagents-typescript-quickstart- Node.js 22 or later
- API key for your chosen provider (Anthropic, OpenAI, or Google)
- `.env` file to store the provider API key (will be gitignored)
How to use deepagents-typescript-quickstart
- 1.Run the install command to add the skill
- 2.Choose a provider and model (e.g., `anthropic:claude-sonnet-5`)
- 3.The skill creates a new `deep-agent/` directory with scaffolded code
- 4.Install dependencies from the generated package.json (excluding Tavily)
- 5.Run the research example with a query like 'What is LangGraph?'
- 6.Review the output and consult the docs for customization options
Use cases
- Quickly prototype a research agent locally without external search API setup
- Experiment with Deep Agents using your preferred LLM provider
- Build a minimal agent scaffold as a starting point for custom extensions
- Test provider-native web search capabilities in an agentic context
- Learn Deep Agents architecture by running the official quickstart
- TypeScript developers exploring Deep Agents
- Teams wanting to prototype agents with minimal setup
- Developers familiar with LangChain who want a quick local agent scaffold
- Anyone testing provider-native search integration
deepagents-typescript-quickstart FAQ
No. This skill uses your chosen provider's built-in web search instead, so you only need that provider's API key.
Any model available through Anthropic, OpenAI, or Google. Pass it as `provider:model` (e.g., `openai:gpt-4`, `google-genai:gemini-3.5-flash`). Default is `anthropic:claude-sonnet-5`.
Yes. The skill creates a standalone project directory. After running the example, consult LangChain's deep-agents-core docs and Managed Deep Agents guide for customization.
The skill skips tracing by default to keep setup minimal. You can add it manually after scaffolding if needed.
No. The skill creates a new `deep-agent/` directory and does all work there, leaving your project untouched.
Full instructions (SKILL.md)
Source of truth, from langchain-ai/langchain-skills.
name: deepagents-typescript-quickstart description: "Scaffold a minimal local Deep Agent in TypeScript by following the official quickstart, using provider-native web search instead of Tavily. Use when the user wants to quickly build or try a Deep Agent locally."
Deep Agents TypeScript quickstart
Follow the live docs — do not invent an alternate API from memory:
https://docs.langchain.com/oss/javascript/deepagents/quickstart
Fetch that page (Docs MCP or HTTP) and implement the research-agent shape it shows (createDeepAgent, research system prompt, invoke with a research question like “What is LangGraph?”). Requires Node 22+.
Local setup constraints
Apply these on top of the quickstart (they keep setup minimal and model-agnostic):
-
Ask which provider/model to use. Showcase that Deep Agents are model-agnostic. Suggested prompt:
Which model should this agent use? Pass a
provider:modelstring — e.g.openai:gpt-5.5,anthropic:claude-sonnet-5,google-genai:gemini-3.5-flash. Default if you're unsure:anthropic:claude-sonnet-5.
We'll use that provider's built-in web search (no separate search API key). -
Create a new directory (e.g.
deep-agent/) and do all work there — do not pollute the open project. -
Do not use Tavily (or
@langchain/tavily). Replace the quickstart's search tool with the chosen provider's built-in web search. Look up the current export/tool shape on that provider's LangChain docs (examples as of writing — re-check if needed):Provider Built-in search tool Anthropic @langchain/anthropictools.webSearch_*()(or equivalent dict)OpenAI { type: "web_search" }Google { google_search: {} }Prefer Anthropic / OpenAI / Google so provider search is available. Only secret: that provider's API key in
.env(gitignored). Skip LangSmith tracing unless they ask. -
Install packages from the quickstart minus Tavily; add the provider package for their model.
-
Run the research example, show output, then stop. Point to
deep-agents-core/ customization / Managed Deep Agents for next steps.
Related skills
More from langchain-ai/langchain-skills and the wider catalog.
ecosystem-primer
Required starting point for LangChain/LangGraph/Deep Agents projects—choose your framework and set up observability.
eval-engineering
Design, build, and audit agent evaluation tasks with structured specs, environments, and verifiers.
framework-selection
Choose the right LangChain/LangGraph/Deep Agents framework layer before writing agent code.
langchain-dependencies
Manage LangChain ecosystem package versions, dependencies, and environment setup for Python and TypeScript projects.
langchain-fundamentals
Build production LangChain agents with create_agent(), tools, and middleware for human-in-the-loop control.
langchain-middleware
Human-in-the-loop approval and custom middleware for LangChain agents with structured output support.