Knowledge & Open Capability

Presets

A preset is a reusable conversation setup. It combines the agents, models, Skills, knowledge bases, prompt requirements, and scenarios you use together, so a new conversation can start with the full configuration in one click.

Open Presets from the sidebar, or click Use preset above the composer on the home screen.

Preset library

Create a preset

  1. Open Presets and click New.
  2. Add a name, description, and avatar so the preset is easy to recognize.
  3. Choose the primary agent and candidate models. Allow fallback when you want another available execution path to take over.
  4. Attach the Skills and knowledge bases this workflow normally needs.
  5. Write the prompt requirements: role, output format, working constraints, and quality bar.
  6. Add audience and scenario tags so the preset is easy to find later.
  7. Save it, return to the home screen, and click Use preset to apply it.

Three sources

  • Builtin presets ship with NomiFun. Their content is read-only, but you can use or duplicate them.
  • Extension presets come from installed extensions and are maintained by the extension.
  • Custom presets are yours to edit, duplicate, enable, disable, or delete.

How presets relate to Skills

Presets decide which capabilities belong together; Skills teach an agent how to perform a workflow. The same Skill can be reused by many presets, so you do not need to duplicate its content.

If you still need a Skill, open Skills from the sidebar:

  • manage local capabilities under Installed Skills;
  • browse ClawHub and SkillHub rankings under Skill Market;
  • inspect the details, then hand a reviewed installation draft to Nomi.

Practical tips

  • Name presets after real workflows—such as “Code review,” “Weekly report,” or “Requirement breakdown”—instead of vague roles.
  • Keep stable rules in the preset and put the changing goal in the conversation composer.
  • Duplicate a builtin preset before customizing so the original remains available as a reference.
  • For long tasks, add fallback models so a temporary provider outage does not stop the workflow.

Source code and release history → GitHub