# Open sourcing model bringup skill

**URL:** <https://forum.modular.com/t/open-sourcing-model-bringup-skill/3203>\
**Category:** Official Announcements\
**Created:** [June 6, 2026, 5:30pm UTC](https://forum.modular.com/t/open-sourcing-model-bringup-skill/3203 "2026-06-06T17:30:12Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Ehsan](https://sea1.discourse-cdn.com/flex001/user_avatar/forum.modular.com/ehsan/32/142_2.png) [@Ehsan](https://forum.modular.com/u/Ehsan)\
**Post date:** [June 6, 2026, 5:30pm UTC](https://forum.modular.com/t/open-sourcing-model-bringup-skill/3203/1 "2026-06-06T17:30:12Z")

</div>

We’ve just open sourced the [import-model](https://github.com/modular/skills/tree/main/import-model) skill which we’ve used a lot for model bringup internally.

### Quick installation

```bash
npx skills add modular/skills --skill import-model

```

then you can use it in Claude Code with `/import-model JetBrains/Mellum2-12B-A2.5B-Thinking to MAX` , Cursor `@import-model` or your favourite agentic tool referring to the skill. I highly recommend to use the **Plan mode** before execution so the agent runs the available scripts such as `inspect` / `check-walls` / `list-archs`, produces a donor choice + delta list + verification gates, then writes code.

Worked example [GitHub - ehsanmok/mellum2-max: Agentic port of JetBrains/Mellum2-12B-A2.5B-Thinking to Modular's MAX · GitHub](https://github.com/ehsanmok/mellum2-max) where I ported Mellum2-12B-A2.5B-Thinking to MAX with 7/7 top-1 token + 271/271 char greedy parity vs the HuggingFace implementation.

Stay tuned for more skills to come. Please let us know your feedback, comments or any issue.