Reusable instructions your team can call - with nothing to escape

Write it once, call it forever

Your organization's routine - how to draft a reply, how to file a report, how to phrase a policy - now travels as a command: your people type /translate-like instructions in the chat they already use, and the assistant follows YOUR playbook. The instruction was written once, in plain language, by whoever knows it best.

Nothing runs, nothing connects

A skill is a template of words, not a piece of code: nothing gets executed, nothing opens a connection. The machine sees the person's words with sensitive details already masked, and the reply it drafts arrives as a clearly-marked draft for the reader - a rulebook your staff can use without your security team losing sleep. There is no way in through a skill: it can add instructions, never doors - a property the platform pins with tests, so you do not need our word for it. You can verify it: the code is inspectable, permission by permission.

Three for everybody, unlimited for the large

The Community edition carries a freemium boundary: three custom skills - the ones your team actually uses daily. The ones the organization writes for everybody: unlimited. The limit is visible in the management page, it explains itself, and the way out is in your hands: delete one of your own to free the slot - no support ticket, no negotiation.

Your rules, not our cleverness

A wrong or forgotten command behaves like any normal message: nothing gets hidden, nothing misfires silently. The server checks every call against your workspace's rules - the command cannot reach material the caller may not see.

See it live

Richiedi una dimostrazione - bring the routine your team repeats every week and watch it become a command:

• Request a demo
• The operator guide(/en/docs/custom-skills)
• The engineering deep dive(/en/blog/custom-skill-authoring)

Facts in this post - FACT-CENSUS rows F-SKILL-1 (cited in the plan). Provenance: the product docs at v0.28.1.