Add skills to Managed Deep Agents

Skills package task-specific procedures and supporting files into reusable directories. MDA discovers them automatically. The agent loads a skill’s full contents only when the task matches the description in the frontmatter.

Managed Deep Agents is in public beta and available on LangSmith Cloud in the US region only.

Put each skill under skills/ at the project root:

my-agent/
  agent.ts
  skills/
    research/
      SKILL.md

For the full project layout, see Project structure.

Add a skill

Use skills for procedures the agent should follow only when a task matches the description in the frontmatter:

Create a skill directory

Each skill directory needs a SKILL.md file with name and description frontmatter:

skills/research/SKILL.md

---
name: research
description: Gather and synthesize context before answering complex questions.
---

# Research

Use this skill when a task needs more than a direct answer.

1. Identify what information is missing.
2. Use `query_db` to look up relevant records.
3. Summarize findings before responding to the user.

For skill authoring patterns and the complete format, see Skills.To understand progressive disclosure, see How the agent uses skills.

Add supporting files (Optional)

A skill directory can also contain supporting scripts, reference files, and templates. Reference these files from SKILL.md so the agent knows when to use them:

skills/
  research/
    SKILL.md
    templates/
      report.md
    scripts/
      fetch_sources.ts

How the agent uses skills

At startup, the agent sees each skill’s name and description. When a task matches a skill’s description, the agent reads the full SKILL.md and follows its instructions. Supporting files are loaded only when needed. The agent cannot modify skills at runtime.

Deployment

When you run mda deploy, MDA syncs every UTF-8 file under skills/ to the agent’s Context Hub. You can then edit skills in the LangSmith UI and apply those changes to the agent. It is best to keep the skill files in the repo as the source of truth for lasting changes, as later deployments sync the project copies again and remove deployed skill files that do not exist locally. For what syncs, what does not, and how to open the repo from a deployment, see Context Hub.

When to use skills

Concept	Role	Loaded when
Instructions	Always-on system prompt	Every run
Skills	Task-specific procedures	When the agent selects them
Memory	Knowledge the agent can update	When durable memory is enabled

For more information, see Project structure.

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

Edit this page on GitHub or file an issue.

Get started

Agent capabilities

Build and deploy

Add skills to Managed Deep Agents

Add a skill

How the agent uses skills

Deployment

When to use skills

​Add a skill

​How the agent uses skills

​Deployment

​When to use skills

Add a skill

How the agent uses skills

Deployment

When to use skills