# databricks-model-serving

A skill that steers the agent through Databricks Model Serving endpoint work: creation, version swaps, traffic splits, logs. Useful when model rollouts go through CLI or MLflow.

- skill · `type=skill` · Claude Code, Codex, Cursor, GitHub Copilot, Gemini CLI, OpenCode · 715 installs · ★ 368
- Author: databricks
- License: LicenseRef-Unknown · policy=link-only · description only; the file stays with the author
- Source: https://github.com/databricks/databricks-agent-skills/tree/HEAD/skills/databricks-model-serving/SKILL.md
- Updated: 2026-10-24
- HTML: https://affpapa.org/en/nevibecode/skills/databricks-model-serving

## What it does

The skill spells out how an endpoint is put together: served entities plus traffic routes in the config, the AI Gateway layer, and the readiness state. From that the agent knows how to build an endpoint for pay-per-token foundation models, provisioned throughput, or a custom MLflow model, and how to poll until `READY`.

Version rollover gets its own section. The alias in Unity Catalog moves to the new number first; then the endpoint config is updated so its `served_entities` and `traffic_config` match. Do only one of the two and the route quietly points at the old version. A table of endpoint types sits alongside, and one blunt rule: check `databricks serving-endpoints -h` rather than guessing at syntax.

Scope lines are drawn hard. Legacy AI Gateway settings on an endpoint belong here, Unity Gateway work goes to `databricks-unity-gateway`, training and hand-written PyFunc to `databricks-ml-training`. The databricks CLI must be at least v0.294.0 and a profile has to be selected, or the first command fails.

## Requirements
- Works with: Databricks, Unity

## Install

### Claude Code (native platform)

1. **Skills CLI (Vercel)**
```bash
npx skills add databricks/databricks-agent-skills --skill databricks-model-serving -a claude-code
```

2. **Manually: put the files in place**
- File: `~/.claude/skills/databricks-model-serving/SKILL.md` — Copy the whole skill folder; for one project use .claude/skills/databricks-model-serving/

### Codex

1. **Skills CLI (Vercel)**
```bash
npx skills add databricks/databricks-agent-skills --skill databricks-model-serving -a codex
```

2. **Manually: put the files in place**
- File: `.agents/skills/databricks-model-serving/SKILL.md` — For all projects use ~/.agents/skills/

### Cursor

1. **Skills CLI (Vercel)**
```bash
npx skills add databricks/databricks-agent-skills --skill databricks-model-serving -a cursor
```

2. **Manually: put the files in place**
- File: `.cursor/skills/databricks-model-serving/SKILL.md` — Cursor also reads .agents/skills/ and .claude/skills/

### GitHub Copilot

1. **GitHub CLI**
```bash
gh skill install databricks/databricks-agent-skills databricks-model-serving
```

2. **Manually: put the files in place**
- File: `.github/skills/databricks-model-serving/SKILL.md`

### Gemini CLI

1. **Skills CLI (Vercel)**
```bash
npx skills add databricks/databricks-agent-skills --skill databricks-model-serving -a gemini-cli
```

2. **Platform CLI**
```bash
gemini skills install https://github.com/databricks/databricks-agent-skills --path skills/databricks-model-serving
```

### Windsurf (Devin Desktop)

**Manually: put the files in place**
- File: `.devin/skills/databricks-model-serving/SKILL.md` — The legacy .windsurf/skills/ path still works

### OpenCode

**Skills CLI (Vercel)**
```bash
npx skills add databricks/databricks-agent-skills --skill databricks-model-serving -a opencode
```

### Claude.ai

**Claude.ai**
- Zip the folder and upload it in Settings → Capabilities → Skills

Where a file is given instead of a command, copy the files from the source to that path: https://github.com/databricks/databricks-agent-skills/tree/HEAD/skills/databricks-model-serving/SKILL.md

## When to use it
- create a Model Serving endpoint from the CLI or MLflow Deployments
- point a production alias at a new model version with no downtime
- split traffic by percentage across two versions for an A/B test
- fetch an endpoint's OpenAPI schema and read its logs and metrics
- set legacy AI Gateway rate limits on an endpoint

## Example prompts
- Run `databricks serving-endpoints create -h` and build the JSON for a new endpoint on a `system.ai` model from what it shows.
- Move the `prod` alias to version 5, update the endpoint config, then confirm `state.ready`.
- Send 90 percent of traffic to the current version and 10 to the new one, and tell me how to roll back.

## Similar
- [skill-creator](https://affpapa.org/en/nevibecode/skills/skill-creator.md) — skill · 414180 installs
- [writing-for-agents](https://affpapa.org/en/nevibecode/skills/writing-for-agents.md) — skill · 466719 installs
- [media-use](https://affpapa.org/en/nevibecode/skills/media-use.md) — skill · 1100000 installs
- [mcp-builder](https://affpapa.org/en/nevibecode/skills/mcp-builder.md) — skill · 130949 installs
- [subagent-driven-development](https://affpapa.org/en/nevibecode/skills/subagent-driven-development.md) — skill · 237746 installs
- [orchestration](https://affpapa.org/en/nevibecode/skills/orchestration.md) — skill · 349800 installs

From the same set: [databricks/databricks-agent-skills](https://affpapa.org/en/nevibecode/by/databricks/databricks-agent-skills.md) — 18 cards

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md