Teach agents how your team works
A skill in Definable is a written playbook an agent reads when a task matches, like how your team handles a refund or writes a release note: instructions, and sometimes scripts, in a folder. Add one from the open catalog, upload or write your own, or ask Builder to draft it. The agent's owner approves every skill.
How skills work
A skill is a written playbook an agent loads when a task matches: how your team handles a refund, writes a release note or escalates a payout.
At a glance
- 5 security auditors
- Upload, write, GitHub or Builder
- Owner approval per agent
- Up to 500 files, 20 MB a skill
- 01
Pick from the open catalog
Browse open skills that five security auditors check. Each one shows Audited, Warn, Fail or Not audited; a Warn or Fail is a finding to read, not a block.
- 02
Or make your own
Upload a folder, a .zip or one SKILL.md, write one in the editor, copy one from a public GitHub repo, or describe it and Builder drafts it.
- 03
The owner says yes
Each agent has an owner who approves its skills. When an admin or Builder adds one, it becomes a request the owner approves or denies.
- 04
Frozen when you publish
A published agent keeps the exact version it was approved with; a newer one waits for the owner. A skill's scripts run only when the agent's sandbox is on.
Skills FAQ
What are skills?
A skill teaches an agent how your team does a task: instructions, and sometimes scripts, in a folder. Add one from the open catalog at skills.sh, upload your own, write one in the editor, copy one from a public GitHub repo or ask Builder to draft it. Catalog skills show what five security auditors found, and the agent's owner approves every skill.
What is Builder?
The agent that builds and fixes your agents. It asks when something is unclear, with the answer it recommends, drafts the agent part by part and tries it before you publish. It never publishes on its own. Builder is on every plan, Free included.
What is the sandbox?
Each conversation gets its own Linux computer where the agent can run code and commands. It's kept between messages, stops after 15 minutes idle and is deleted 30 days after it was last used. Secrets you add become environment variables, hidden in the output, and only your team's own chats and schedules get them.