Deploy with the CLI

Deploy a model-agnostic, open source agent to production with the Deep Agents CLI

Deep Agents Deploy takes your agent configuration and deploys it as a LangSmith Deployment: a horizontally scalable server with 30+ endpoints including MCP, A2A, Agent Protocol, human-in-the-loop, and memory APIs. Built on open standards:

• Open source harness: MIT licensed, available for Python and TypeScript
• AGENTS.md: open standard for agent instructions
• Agent Skills: open standard for agent knowledge and actions
• Any model, any sandbox: no provider lock-in
• Open protocols: MCP, A2A, Agent Protocol
• Self-hostable: LangSmith Deployments can be self-hosted so memory stays in your infrastructure

Deep Agents Deploy is currently in beta. APIs, configuration format, and behavior may change between releases. See the releases page for detailed changelogs.

Compare to Claude Managed Agents

	Deep Agents Deploy	Claude Managed Agents
Model support	OpenAI, Anthropic, Google, Bedrock, Azure, Fireworks, Baseten, OpenRouter, many more	Anthropic only
Harness	Open source (MIT)	Proprietary, closed source
Sandbox	LangSmith, Daytona, Modal, Runloop, or custom	Built in
MCP support	✅	✅
Skill support	✅	✅
AGENTS.md support	✅	❌
Agent endpoints	MCP, A2A, Agent Protocol	Proprietary
Self hosting	✅	❌

What you’re deploying

deepagents deploy packages your agent configuration and deploys it as a LangSmith Deployment. You configure your agent with a few parameters:

Parameter	Description
model	The LLM to use. Any provider works — see supported models.
AGENTS.md	The system prompt, loaded at the start of each session.
skills	Agent Skills for specialized knowledge and actions. Skills are synced into the sandbox so the agent can execute them at runtime. See skills docs.
user/	Per-user writable memory. If a AGENTS.md template is present in the user folder, the agents seeds the template per user (if the folder is empty the agents creates an empty AGENTS.md). Writable at runtime. Preloaded into the agent’s context via the memory middleware.
mcp.json	MCP tools (HTTP/SSE). See MCP docs.
subagents/	Specialized subagents the main agent can delegate to. Each subdirectory contains its own deepagents.toml, AGENTS.md, and optionally a skills folder. See subagents.
sandbox	Optional execution environment. Thread-scoped sandboxes are provisioned per thread and will be re-created if the server restarts. Use scope = "assistant" if you need sandbox state that persists across threads. See sandbox providers.

Install

Install the CLI or run directly with uvx:

uv

uv tool install deepagents-cli

Usage

deepagents init [name] [--force]                                             # scaffold a new project
deepagents dev  [--config deepagents.toml] [--port 2024] [--allow-blocking]  # bundle and run locally
deepagents deploy [--config deepagents.toml] [--dry-run]                     # bundle and deploy

By default, deepagents deploy looks for deepagents.toml in the current directory. Pass --config to use a different path:

deepagents deploy --config path/to/deepagents.toml

deepagents deploy fully rebuils and creates a new revision on every invocation. Use deepagents dev for local iteration.

deepagents init

Scaffold a new agent project:

deepagents init my-agent

This creates the following files:

File	Purpose
deepagents.toml	Agent config — name, model, optional sandbox
AGENTS.md	System prompt loaded at session start
.env	API key template (GOOGLE_API_KEY, LANGSMITH_API_KEY, etc.)
mcp.json	MCP server configuration (empty by default)
skills/	Directory for Agent Skills, with an example review skill

After init, edit AGENTS.md with your agent’s instructions and run deepagents deploy. Optionally add a user/ directory with per-user memory templates — see User Memory.

Project layout

The deploy command uses a convention-based project layout. Place the following files alongside your deepagents.toml and they are automatically discovered:

my-agent/
├── deepagents.toml
├── AGENTS.md
├── .env
├── mcp.json
├── skills/
│   ├── code-review/
│   │   └── SKILL.md
│   └── data-analysis/
│       └── SKILL.md
├── subagents/
│   └── researcher/
│       ├── deepagents.toml
│       └── AGENTS.md
└── user/
    └── AGENTS.md

File/directory	Purpose	Required
AGENTS.md	Memory for the agent. Provides persistent context (project conventions, instructions, preferences) that is always loaded at startup. Read-only at runtime.	Yes
skills/	Directory of skill definitions. Each subdirectory should contain a SKILL.md file. Read-only at runtime.	No
user/	Per-user writable memory. If a AGENTS.md template is present in the user folder, the agents seeds the template per user (if the folder is empty the agents creates an empty AGENTS.md). Writable at runtime. Preloaded into the agent’s context via the memory middleware.	No
subagents/	Subagents the main agent can delegate to. Each subdirectory must contain a deepagents.toml, AGENTS.md, and optionally a skills folder. Auto-discovered at bundle time.	No
mcp.json	MCP server configuration. Only http and sse transports are supported in deployed contexts.	No
.env	Environment variables (API keys, secrets). Placed alongside deepagents.toml at the project root.	No

mcp.json must only contain servers using http or sse transports. Servers using stdio transport are not supported in deployed environments because there is no local process to spawn.

Convert stdio servers to HTTP or SSE before deploying.

Configuration file

deepagents.toml configures the agent’s identity and sandbox environment. Only the [agent] section is required. The [sandbox] section is optional and defaults to no sandbox.

[agent]

(Required)

Core agent identity. For more on model selection and provider configuration, see supported models.

name · string · required

Name for the deployed agent. Used as the assistant identifier in LangSmith.

model · string

Model identifier in provider:model format. See supported models.

deepagents.toml

[agent]
name = "research-assistant"
model = "google_genai:gemini-3.1-pro-preview"

The name field is the only required value in the entire configuration file. Everything else has defaults.

Skills, user memories, subagents, MCP servers, and model dependencies are auto-detected from the project layout:

• Skills: the bundler recursively scans skills/, skipping hidden dotfiles, and bundles the rest.
• User memory: if user/ exists, a single AGENTS.md is bundled as per-user memory (from user/AGENTS.md if present, otherwise empty). At runtime, each user gets their own copy (seeded on first access, never overwritten). The agent can read from and write to this file.
• Subagents: if subagents/ exists, the bundler scans for valid subdirectories (each must contain deepagents.toml and AGENTS.md). The main agent receives a task tool to delegate work to each subagent by name. See subagents.
• MCP servers: if mcp.json exists, it is included in the deployment and langchain-mcp-adapters is added as a dependency. Only HTTP/SSE transports are supported (stdio is rejected at bundle time).
• Model dependencies: the provider: prefix in the model field determines the required langchain-* package (e.g., google_genai -> langchain-google-genai). This includes models specified in subagent configs.
• Sandbox dependencies: the [sandbox].provider value maps to its partner package (e.g., daytona -> langchain-daytona).

[sandbox]

Configure the isolated execution environment where the agent runs code. Sandboxes provide a container with a filesystem and shell access, so untrusted code cannot affect the host. For supported providers and advanced sandbox configuration, see sandboxes.

When omitted or set to provider = "none", the sandbox is disabled. Sandboxes are for if you need code execution or skill script execution.

provider · string · default: "none"

Sandbox provider. Determines where the container runs. Supported values: "none", "daytona", "modal", "runloop", "langsmith" (private beta). See sandbox integrations for provider details.

template · string · default: "deepagents-deploy"

Provider-specific template name for the sandbox environment.

image · string · default: "python:3"

Base Docker image for the sandbox container.

scope · string · default: "thread"

Sandbox lifecycle scope. "thread" creates one sandbox per conversation. "assistant" shares a single sandbox across all conversations for the same assistant.

Scope behavior:

• "thread" (default): Each conversation gets its own sandbox. Different threads get different sandboxes, but the same thread reuses its sandbox across turns. Use this when each conversation should start with a clean environment.
• "assistant": All conversations share one sandbox. Files, installed packages, and other state persist across conversations. Use this when the agent maintains a long-lived workspace like a cloned repo.

[auth]

Optionally, add an [auth] section to enable user authentication on the deployed agent. When present, the bundler generates an auth.py file that validates Bearer tokens, scopes all resources per authenticated user, and wires everything into the deployment automatically.

provider · string · required

Auth provider. Supported values: "supabase", "clerk".

deepagents.toml

[agent]
name = "my-agent"
model = "google-genai:gemini-3.1-pro-preview"

# Optional — remove to deploy without auth
[auth]
provider = "supabase"   # supabase | clerk

The required environment variables depend on the provider:

Provider	Required env vars
supabase	SUPABASE_URL, SUPABASE_PUBLISHABLE_DEFAULT_KEY
clerk	CLERK_SECRET_KEY

Add these to your .env alongside your other credentials. The bundler validates that all required vars are present at deploy time and fails fast with a clear error if any are missing.

Runtime behavior:

• Unauthenticated requests return 401.
• On success, the authenticated user’s identity is injected into config.configurable.langgraph_auth_user_id.
• All resources (threads, runs, store) are automatically scoped per user via metadata.owner.
• LangSmith Studio bypasses auth for local development.

.env

Place a .env file alongside deepagents.toml with your API keys:

# Required — model provider keys
ANTHROPIC_API_KEY=sk-...
OPENAI_API_KEY=sk-...
# ...etc.

# Required for deploy and LangSmith sandbox
LANGSMITH_API_KEY=lsv2_...

# Optional — sandbox provider keys
DAYTONA_API_KEY=...
MODAL_TOKEN_ID=...
MODAL_TOKEN_SECRET=...
RUNLOOP_API_KEY=...

# Required if [auth] provider = "supabase"
SUPABASE_URL=https://your-project.supabase.co
SUPABASE_PUBLISHABLE_DEFAULT_KEY=eyJhbGc...

# Required if [auth] provider = "clerk"
CLERK_SECRET_KEY=sk_test_...

Authentication

When [auth] is configured in deepagents.toml, pass the token from your auth provider in the Authorization header:

curl

curl -X POST https://your-deployment-url/threads \
  -H "Authorization: Bearer YOUR_ACCESS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"metadata": {}}'

Python (langgraph-sdk)

from langgraph_sdk import get_client

client = get_client(
    url="https://your-deployment-url",
    headers={"Authorization": "Bearer YOUR_ACCESS_TOKEN"},
)

thread = await client.threads.create()

Provider	Where to get the token
Supabase	Supabase session access_token from supabase.auth.getSession()
Clerk	Clerk session token from getToken()

Each user’s threads and memory are isolated automatically—User B cannot see User A’s threads.

Sandbox providers

Set [sandbox].provider in deepagents.toml and add the required env vars to .env. For available providers, see sandbox integrations. For lifecycle patterns and SDK usage, see sandboxes.

Deployment endpoints

The deployed server exposes:

• MCP: call your agent as a tool from other agents
• A2A: multi-agent orchestration via A2A protocol
• Agent Protocol: standard API for building UIs
• Human-in-the-loop: approval gates for sensitive actions
• Memory: short-term and long-term memory access

Examples

A content writing agent with per-user preferences that the agent can update:

deepagents.toml

[agent]
name = "deepagents-deploy-content-writer"
model = "google_genai:gemini-3.1-pro-preview"

my-content-writer/
├── deepagents.toml
├── AGENTS.md
├── skills/
│   ├── blog-post/SKILL.md
│   └── social-media/SKILL.md
└── user/
    └── AGENTS.md          # writable — agent learns user preferences

A coding agent with a LangSmith sandbox for running code:

deepagents.toml

[agent]
name = "deepagents-deploy-coding-agent"
model = "google_genai:gemini-3.1-pro-preview"

[sandbox]
provider = "langsmith"
template = "coding-agent"
image = "python:3.12"

A GTM strategy agent that delegates research to a subagent:

my-gtm-agent/
├── deepagents.toml
├── AGENTS.md
├── skills/
│   └── competitor-analysis/
│       └── SKILL.md
└── subagents/
    └── market-researcher/
        ├── deepagents.toml
        ├── AGENTS.md
        └── skills/
            └── analyze-market/
                └── SKILL.md

User Memory

User memory gives each user their own writable AGENTS.md that persists across conversations. To enable it, create a user/ directory at your project root:

user/
└── AGENTS.md          # optional — seeded as empty if not provided

If the user/ directory exists (even if empty), every user gets their own AGENTS.md at /memories/user/AGENTS.md. If you provide user/AGENTS.md, its contents are used as the initial template; otherwise an empty file is seeded.

At runtime, user memory is scoped per user via custom auth (runtime.server_info.user.identity). The first time a user interacts with the agent, their namespace is seeded with the template. Subsequent interactions reuse the existing file — the agent’s edits persist, and redeployments never overwrite user data.

How it works

1. Bundle time — the bundler reads user/AGENTS.md (or uses an empty string) and includes it in the seed payload.
2. Runtime (first access) — when the agent sees a user_id for the first time, it writes the AGENTS.md template to the store under that user’s namespace. Existing entries are never overwritten.
3. Preloaded — the user AGENTS.md is passed to the memory middleware, so the agent sees its contents in context at the start of every conversation.
4. Writable — the agent can update it using the edit_file tool. The shared AGENTS.md file and skills folder are read-only.

Permissions

Path	Writable	Scope
/memories/AGENTS.md	No	Shared (assistant-scoped)
/memories/skills/**	No	Shared (assistant-scoped)
/memories/user/**	Yes	Per-user (user_id-scoped)
/memories/subagents/<name>/**	By subagent only	Per-subagent (isolated)

User identity

The user_id is resolved from custom auth via runtime.user.identity. The platform injects the authenticated user’s identity automatically — no need to pass it through configurable. If no authenticated user is present, user memory features are gracefully skipped for that invocation.

Subagents

Subagents let the main agent delegate specialized tasks to isolated child agents. Each subagent has its own system prompt, optional skills, and optional MCP tools. The main agent receives a task tool that dispatches work to subagents by name.

For background on why subagents are useful and how they work at the SDK level, see Subagents.

Directory structure

Create a subagents/ directory at your project root. Each subdirectory is a subagent:

my-agent/
├── deepagents.toml
├── AGENTS.md
└── subagents/
    ├── researcher/
    │   ├── deepagents.toml       # name, description, optional model override
    │   ├── AGENTS.md             # subagent system prompt
    │   ├── skills/               # optional — subagent-specific skills
    │   │   └── analyze-market/
    │   │       └── SKILL.md
    │   └── mcp.json              # optional — HTTP/SSE MCP tools
    └── writer/
        ├── deepagents.toml
        └── AGENTS.md

Each subagent subdirectory must contain:

File	Purpose
deepagents.toml	Subagent config with [agent].name and [agent].description
AGENTS.md	System prompt for the subagent

Each subagent subdirectory may contain:

File	Purpose
skills/	Subagent-specific skills (with SKILL.md files)
mcp.json	MCP server configuration (HTTP/SSE only; stdio is rejected)

Subagent configuration

name · string · required

Unique identifier for the subagent. Must be unique across all subagents.

description · string · required

What this subagent does. The main agent uses this to decide when to delegate. Must be non-empty.

model · string

Model override in provider:model format. Omit to inherit the main agent’s model.

subagents/researcher/deepagents.toml

[agent]
name = "researcher"
description = "Researches market trends, competitors, and target audiences"
model = "google_genai:gemini-3.1-pro-preview"

Inheritance

Subagents inherit some properties from the main agent by default:

Property	Inherited	Notes
Model	Yes	Override with model in the subagent’s deepagents.toml
Tools	Yes	Override by adding mcp.json to the subagent directory
Skills	No	Declare explicitly in the subagent’s own skills/ directory

Memory isolation

Each subagent gets a dedicated, isolated memory namespace at /memories/subagents/<name>/. The subagent’s AGENTS.md and skills are seeded into this namespace at deploy time.

Path	Main agent	Subagent
/memories/AGENTS.md	Read	No access
/memories/skills/**	Read	No access
/memories/user/**	Read + Write	No access
/memories/subagents/<name>/**	Read	Read + Write

Example

A go-to-market agent that delegates research to a specialized subagent:

deepagents.toml

[agent]
name = "gtm-strategist"
model = "google_genai:gemini-3.1-pro-preview"

subagents/researcher/deepagents.toml

[agent]
name = "researcher"
description = "Researches market trends, competitors, and target audiences to inform GTM strategy"
model = "google_genai:gemini-3.1-pro-preview"

subagents/researcher/AGENTS.md

# Market Researcher

You are a market research specialist. Your job is to gather and synthesize
market data to support go-to-market decisions.

## Focus Areas
- Market sizing: TAM, SAM, SOM estimates
- Competitor analysis: product positioning, pricing, market share
- Audience segmentation: demographics, psychographics, buying behavior

Limitations

• MCP: HTTP/SSE only. Stdio transports are rejected at bundle time.
• No custom Python tools. Use MCP servers to expose custom tool logic.

---

Edit this page on GitHub or file an issue.

Connect these docs to Claude, VSCode, and more via MCP for real-time answers.