skills.rest
flemx
flemxCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

managed-deep-agents

Build, test, and deploy code-first Deep Agents to LangSmith with the mda CLI.

Updated Jul 16, 2026
Search Tags:#langgraph#agent deployment#deep agents#langsmith#mcp connectors#mda cli#cron schedules
One-click install
npx skills add https://github.com/flemx/salesforce-langgraph-agent --skill managed-deep-agents-flemx
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: managed-deep-agents
Source: https://github.com/flemx/salesforce-langgraph-agent/tree/main/.agents/skills/managed-deep-agents
Command: npx skills add https://github.com/flemx/salesforce-langgraph-agent --skill managed-deep-agents-flemx

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires managed-deepagents.

What problem does it solve? Deploying a custom agent usually means standing up your own server, wiring persistence, memory, and scheduling by hand. This Skill guides you through Managed Deep Agents, a hosted LangSmith runtime where an agent is just a directory of code that the mda CLI compiles and deploys for you. ## Core Features & Use Cases - Code-first agent definition: Author agents in Python or TypeScript with define_deep_agent / defineDeepAgent, plus authored tools, middleware, and MCP connectors. - Managed runtime resources: Configure cron schedules, LangSmith sandboxes, deploy-owned skills, and Context Hub-synced instructions and memory. - CLI workflow: Scaffold with mda init, test locally in LangSmith Studio with mda dev, and ship with mda deploy, including secrets forwarding and cron reconciliation. - Use Case: You want a research assistant agent with a daily digest cron job and human-in-the-loop approval on sensitive tools, deployed to LangSmith without managing any infrastructure. ## Quick Start Ask the agent to scaffold a new Managed Deep Agent project with mda init, define an agent with a model and tools, and deploy it to LangSmith using mda deploy.

Frequently Asked Questions about managed-deep-agents

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I deploy a Deep Agent to LangSmith?▼

Install the managed-deepagents package, scaffold a project with mda init, define your agent with define_deep_agent or defineDeepAgent, then run mda deploy. The CLI compiles the project, syncs instructions and skills to Context Hub, forwards .env secrets, and waits until the revision is deployed.

How do I test a Managed Deep Agent locally before deploying?▼

Run mda dev from the project root to compile the project into .mda/build and start a local LangGraph dev server in LangSmith Studio. For Python projects, run uv sync first, and set LANGSMITH_API_KEY in your .env or shell.

Does Managed Deep Agents support MCP servers?▼

Yes, declare remote MCP servers in connectors/mcp.py or connectors/mcp.ts with a named mcp export. Only http and sse transports are supported; stdio servers are rejected, and configuration is validated at build or dev startup.

When should I use a standard LangSmith Deployment instead of Managed Deep Agents?▼

Use a standard LangGraph deployment when you need custom application code, custom routes, advanced authentication, stronger isolation, maximum scalability, or a region outside US LangSmith Cloud. Managed Deep Agents is CLI-first, US-only, and does not support self-hosted or Hybrid setups during private beta.

Why does my Managed Deep Agent model fail to resolve?▼

Model identifiers must use the provider-prefixed form like openai:gpt-5.5, not a bare model name. The runtime resolves them with init_chat_model, so any provider supported by init_chat_model works as long as the provider API key is available.

How do I add human-in-the-loop approval to agent tool calls?▼

Set interrupt_on or interruptOn in define_deep_agent with the tools that require approval, optionally combined with permissions. During mda dev you respond in LangSmith Studio; on deployed agents you resume through the LangGraph server API with a Command resume payload.

Related Skills

Explore More Software Engineering Skills→
langchain-ailangchain-ai
1.2k

managed-deep-agents

Build, test, and deploy code-first Deep Agents on LangSmith using the mda CLI.

Official
Advanced
FMurrayFMurray

instrumenting-with-mlflow-tracing

Instruments Python and TypeScript applications with MLflow Tracing for LLM observability.

Community
Intermediate
langchain-ailangchain-ai
1.2k

langgraph-cli

Scaffold, develop, build, and deploy LangGraph applications using the langgraph CLI.

Official
Intermediate
olimpus-softolimpus-soft

arch-py

Analyze Python application architectures for design patterns and best practices.

Official
Advanced