Skip to main content
Skills package task-specific procedures and context into reusable directories. You can define them in markdown files, and they are picked up automatically by the agent.
Managed Deep Agents is in public beta and available on LangSmith Cloud in the US region only.

​
Project structure

Keep the agent entry point at the project root and define each skill under skills/:

​
Add a skill

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

​
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. This progressive disclosure gives the agent access to detailed procedures without adding every skill’s full contents to its context.

​
Syncing to Context Hub

When you run mda deploy, every UTF-8 file under skills/ is automatically synced to the agent’s Context Hub repo. You can then edit skills in the LangSmith UI and make the changes available to the agent. A later deployment syncs the project copies again and removes deployed skill files that no longer exist locally.

​
How skills compare to other concepts

Skills is context that is loaded dynamically, when the agent chooses to. The agent cannot modify them. Use instructions for behavior that should ALWAYS be loaded by the agent. Use memory for knowledge you want the agent to be able to update. For skill authoring patterns and the complete format, see Skills.
Connect these docs to Claude, VSCode, and more via MCP for real-time answers.
Edit this page on GitHub or file an issue.