菜单
Use when Databricks skills need updating, user asks to refresh or sync skills from upstream, or skills seem outdated compared to the ai-dev-kit repo
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | refresh-databricks-skills |
| description | Use when Databricks skills need updating, user asks to refresh or sync skills from upstream, or skills seem outdated compared to the ai-dev-kit repo |
Pulls the latest Databricks skills from the upstream source repo and replaces all existing Databricks skills in the project while preserving non-Databricks skills (e.g., superpowers workflow skills).
Source repo: https://github.com/databricks-solutions/ai-dev-kit (path: databricks-skills/)
Clone the upstream repo (shallow clone for speed):
git clone --depth 1 https://github.com/databricks-solutions/ai-dev-kit.git $TMPDIR/ai-dev-kit
Identify non-Databricks skills to preserve. These are the superpowers workflow skills that live alongside Databricks skills. List them by checking which directories in .claude/skills/ do NOT have a matching folder in the upstream databricks-skills/ directory. Common superpowers skills include: brainstorming, dispatching-parallel-agents, executing-plans, finishing-a-development-branch, receiving-code-review, requesting-code-review, subagent-driven-development, systematic-debugging, test-driven-development, using-git-worktrees, using-superpowers, verification-before-completion, writing-plans, writing-skills. Also preserve any other project-specific skills (like this one: refresh-databricks-skills).
Remove old Databricks skills from .claude/skills/, keeping all non-Databricks skills identified above.
Copy new Databricks skills from the cloned repo. Copy every directory under databricks-skills/ except TEMPLATE:
SKILLS_DIR=".claude/skills"
UPSTREAM="$TMPDIR/ai-dev-kit/databricks-skills"
for dir in "$UPSTREAM"/databricks-* "$UPSTREAM"/spark-*; do
[ -d "$dir" ] && cp -r "$dir" "$SKILLS_DIR/$(basename "$dir")"
done
Clean up the cloned repo:
rm -rf $TMPDIR/ai-dev-kit
Report the count of skills added, removed, and updated.
If the project is deployed as a Databricks App, remind the user to sync the updated skills to the workspace and redeploy:
databricks workspace import-dir <local-path> <workspace-path> --overwrite --profile <profile>
databricks apps deploy <app-name> --source-code-path <workspace-path> --profile <profile>
refresh-databricks-skills must be preserved during the refresh.--depth 1: Full clone is slow and unnecessary. Always shallow clone.This skill should be used when the user asks to "write Gherkin", "create feature files", "generate BDD scenarios", "write acceptance tests in Gherkin", "create Behave features", "write Given When Then tests", "BDD test cases for my pipeline", "Gherkin for Unity Catalog", or wants to translate requirements into Gherkin feature files for Databricks.
This skill should be used when the user asks to "run BDD tests", "execute Behave", "run Gherkin tests", "run my feature files", "behave test results", "run smoke tests", "BDD test report", or needs to execute Behave test suites with specific options like tag filtering, parallel execution, or CI reporting.
This skill should be used when the user asks to "set up BDD", "create a Behave project", "scaffold BDD tests", "initialize Behave", "add BDD to my project", "set up Gherkin testing", "create test structure for Behave", or mentions setting up behavior-driven development testing. Generates a complete Behave project structure wired to Databricks SDK.
This skill should be used when the user asks to "write step definitions", "implement BDD steps", "generate step code", "create Behave steps", "implement Given When Then", "write Python steps for Gherkin", "step definitions for Databricks", or needs to create Python step implementations for existing Gherkin feature files.
Deploy and query Databricks Model Serving endpoints. Use when (1) deploying MLflow models or AI agents to endpoints, (2) creating ChatAgent/ResponsesAgent agents, (3) integrating UC Functions or Vector Search tools, (4) querying deployed endpoints, (5) checking endpoint status. Covers classical ML models, custom pyfunc, and GenAI agents.
Create and manage Databricks Agent Bricks: Knowledge Assistants (KA) for document Q&A, Genie Spaces for SQL exploration, and Supervisor Agents (MAS) for multi-agent orchestration. Use when building conversational AI applications on Databricks.