Use cases

One teammate figures it out. Everyone can run it.

The best way your team does anything usually lives in one person's head — or one person's chat history. Rafter turns it into a skill: taught once, versioned, runnable from every teammate's AI tool.

Where good workflows go to hide

Someone on your team has already worked out how to close the month, onboard a client, or ship a release without drama. The steps exist — in their head, their notes, a chat scrollback nobody else can see.

So the workflow spreads by interruption: ask the person, wait, get a slightly different version of the answer each time. When they're away, everyone else's work gets slower and worse. When they leave, it leaves with them.

Teach it once, run it anywhere

A skill in Rafter is the workflow written down the way your team actually does it — the steps, the gotchas, the reasons behind the rules. Write it in the dashboard, or push it straight from Claude Code the moment you notice you've just done something worth keeping.

From then on it isn't documentation, it's behavior: any teammate's AI tool — Claude, ChatGPT, Cursor, Codex — finds the skill when the task comes up and follows it instead of improvising. Wiki links connect it to the memories and decisions it depends on, so the context travels with it.

When the workflow improves, you edit the skill. Everyone's next run uses the new version, and the version history keeps the old way inspectable.

A month-end-close skill in Rafter: numbered steps, gotchas, and a connections rail showing the memories it cites
A skill with its steps — and the memories it cites, resolved into a graph.

Documentation describes the work. A skill is the work — runnable.

What you’d actually type

Two minutes to connect the tool you already use.From $9.99 per user — see pricing.

Connect your tools