Deep Agents Code

What is Deep Agents Code?

Deep Agents Code, run with the command dcode, is an open-source coding agent for the terminal. It is built on LangChain's Deep Agents SDK and works with any tool-calling model, with OpenAI, Anthropic and Google available out of the box. You install it with one script, connect a provider with the /auth command and describe a task. The agent proposes changes as diffs for your approval, can run shell commands to test code and can search the web through Tavily. Persistent memory keeps project conventions between sessions, skills add custom instructions, and subagents take on tasks in parallel. Goals and rubrics let you define what done means, human-in-the-loop controls gate sensitive tool calls, and LangSmith tracing records runs. There is also a non-interactive mode for scripts. The docs list no price for dcode itself, so the only cost is your own model provider usage.

Use cases

• Code from the terminal with an AI agent
• Delegate tasks to parallel subagents
• Run agent tools in a remote sandbox
• Require approval for risky tool calls
• Trace agent runs in LangSmith
• Automate tasks by piping input to dcode

Who it's for

• Developers
• LangChain users
• AI engineers

Why use Deep Agents Code

• Model choice stays open, so you can switch providers mid-project instead of committing to one vendor.
• Diffs and approval prompts let you review each change before the agent touches your files.
• Memory and skills carry your project conventions from one session to the next, so you repeat yourself less.
• Remote tool execution and LangSmith tracing help teams that want isolation or an audit trail of agent actions.
• Because it is open source and configured through plain files, administrators can enforce settings with a managed config.

Deep Agents Code features

• Works with any tool-calling LLM, with model switching via /model
• Persistent memory across conversations
• Custom skills, invoked with /skill
• Subagents for parallel task execution
• Goals and rubrics to check whether work is done
• Human-in-the-loop approval for sensitive tool operations
• Context compaction that offloads older messages to storage
• MCP tools loaded from Model Context Protocol servers
• LangSmith tracing for observability
• Interactive and non-interactive modes

Deep Agents Code integrations

• LangSmith
• Tavily
• OpenAI
• Anthropic
• Google
• Model Context Protocol

How to get started with Deep Agents Code

1. Run the curl install script from the Deep Agents Code docs, then launch dcode in your terminal.
2. Use /auth to add credentials for OpenAI, Anthropic, Google or another provider.
3. Give the agent a task in plain language and review the diffs it proposes.
4. Approve changes, let it run shell commands to test, and use /remember to save lessons to memory.
5. Optionally add a LangSmith key to trace runs, or pipe input to dcode for scripted use.

Deep Agents Code limitations

• You must supply your own model provider credentials
• Not officially supported on Windows; the docs suggest WSL
• Web search needs a separate Tavily key

Deep Agents Code pricing

Deep Agents Code is open source, so you can use and self-host it for free.

Prices can change and may vary by country or exclude tax. Check the official pricing page.

Deep Agents Code alternatives

The main Deep Agents Code alternatives listed on Debuly are Claude Code, Gemini CLI, Aider, OpenHands, OpenCode and Cursor. The table compares pricing and platforms; order is A–Z.

1. Ai

   Aider

   Open-source AI pair programming tool that runs in your terminal and edits code in your local git repository using the LLM you choose.

   CodingDeveloper platforms
   Open sourceVisit Aider (official site)
2. 

   Claude Code

   Anthropic's coding agent that fixes bugs, writes tests and runs migrations from your terminal, IDE, desktop app, web or Slack.

   CodingAutomation and agentsDeveloper platforms
   PaidVisit Claude Code (official site)
3. 

   Cursor

   AI code editor and coding agent with cloud agents, Bugbot code review and multi-provider models, sold with Grok and X plans.

   CodingAutomation and agentsDeveloper platforms
   FreemiumVisit Cursor (official site)
4. 

   Gemini CLI

   Open-source command-line agent from Google that runs Gemini models in your terminal to read, edit and automate work in local projects.

   CodingAutomation and agentsDeveloper platforms
   Open sourceVisit Gemini CLI (official site)
5. 

   OpenCode

   Open-source AI coding agent for terminal, desktop and IDE, with optional low-cost Go model subscriptions and a team Console

   CodingDeveloper platformsAutomation and agents
   FreemiumVisit OpenCode (official site)
6. 

   OpenHands

   Open-source AI coding agent platform with a free Agent Canvas, a Teams plan, and self-hosted Enterprise control

   CodingAutomation and agentsDeveloper platforms
   FreemiumVisit OpenHands (official site)

Deep Agents Code FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from docs.langchain.com · fetched 23 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.