A skill packages instructions and the material they need
A skill is a folder containing aSKILL.md file, and that file has two parts: YAML frontmatter and markdown instructions.
The frontmatter description is what an agent reads to decide when the skill applies. The markdown body is the procedure it follows once it does. Reference material, scripts and examples sit in the folder alongside it, and none of it loads until the skill is actually used.
Agent Skills is an open standard, so the folder isn’t tied to one tool. That is why work your team has already written moves across intact.
An imported skill keeps its origin visible
Imported skills appear in the playbook editor’s Skill library, grouped by where they came from, next to the skills your connected integrations already provide. Add one to a playbook and it becomes a step in the graph, named for the skill it runs and badged with the system it belongs to. You can read the whole procedure off the canvas and still see which parts are yours. A skill can also be attached to an agent rather than a playbook, which makes it available while the agent works a task.Connect an MCP server instead of rewriting it
The Model Context Protocol is an open standard for connecting AI applications to external systems: data sources, tools and workflows. A server exposes its capabilities once, and any client that speaks the protocol can use them. If you already run one for an internal system, connect that server instead of rebuilding its tools here.We handle the import today
Send your account manager either the skill folder,SKILL.md and supporting files included, or the connection details for your MCP server. We add it to your workspace.
Self-serve import is on the way. Until it lands, every import goes through your account manager.
What’s next
Playbooks
See how skills chain into a repeatable workflow.
Build a playbook
Turn a described procedure into a step graph.