Skills

A skill is a reusable set of instructions you write once and every agent in your workspace can use. Think of it as a playbook: "how we file a bug", "our brand voice for customer replies", "the steps for the weekly revenue rollup".

Instead of repeating the same instructions to an agent every time — or pasting them into one agent's setup where no one else can reuse them — you capture the procedure as a skill. Agents load it only when a task matches, so it never gets in the way the rest of the time.

How agents use skills#

Each skill has a short description that says what it does and when to use it. Every agent sees that one line in its list of available skills. When a task matches, the agent loads the full instructions and follows them — pulling in the right tools and your workspace's specific details (which space, which board, which recipients) as the skill spells out.

You don't have to wire anything up per agent: when you enable a skill, it's available to every agent in the workspace right away.

Creating a skill#

Open Settings → Skills and choose New skill. You'll give it:

• Name — a short identifier, like customer-reply-voice.
• Description — one line on what it does and when to use it. This is the only thing an agent sees until it loads the skill, so make it count.
• When to use (optional) — an extra hint about the trigger.
• Instructions — the full procedure in plain markdown: the steps, which tools to use, the specific details, and when to stop.
• Plugins it uses (optional) — the capabilities the procedure relies on, so the agent loads them automatically.

Leave Enabled on and save. The skill is immediately live for every agent.

Inherited skills#

Monad ships with a set of skills every workspace gets for free — things like setting up a new space well or modelling a database. You'll see these under Inherited skills; they're read-only. If you want to override one for your workspace, create your own skill with the same name and yours takes precedence.

Let an agent write a skill for you#

After @monad (or another agent) does something multi-step you liked, you can say "turn that into a skill the team can reuse". The agent distils what it did into a new skill and saves it — live to everyone immediately, with an agent badge so you can see it came from an agent and review it. You can edit or disable it anytime.

History and reverting#

Every edit is kept. Open a skill's History to see each version — when it was made and whether it came from a person, an agent, or an import — and Revert to restore an earlier version. The revert is saved as a new version, so nothing is lost.

Pausing vs. deleting#

If you just want to take a skill out of rotation for a while, switch off Enabled — agents stop seeing it, but it stays in your list to turn back on later. Delete removes the skill from every agent right away and frees up its name, so you can reuse that name for a new skill.

Bringing skills in and out#

If you've written skills elsewhere (the common SKILL.md format), choose Import SKILL.md and paste it in — Monad reads it leniently and warns you about anything cosmetic. To take a skill with you, use Export to download it as a SKILL.md file.

Who can do what#

Any member can create, edit, enable, and delete skills — it's not limited to admins. Because an enabled skill reaches every agent, the History and the source badges (human / agent / imported) are there so the whole workspace can see what's live, where it came from, and when it last changed.

Tips for good skills#

• Keep the description specific — "Use when drafting any customer-facing message" beats "customer stuff".
• Write the instructions as a clear procedure, not a wall of background.
• Prefer one well-scoped skill over a sprawling one; edit an existing skill rather than creating near-duplicates.