Add Claude Skill for AI Functions

[CEDipEngineering]

Summary

Adds a new databricks-ai-functions skill covering all Databricks built-in AI Functions
with SQL and PySpark examples, function selection guidance, and an end-to-end document
processing pipeline pattern.

Why This Is Useful

Databricks AI Functions (ai_classify, ai_extract, ai_query, ai_forecast, etc.)
are often underused because developers default to ai_query for everything or reach for
external LLM libraries when a simpler built-in function would suffice. This skill gives
Claude Code the context to:

• Pick the right function for the task (task-specific functions first, ai_query only as
  a last resort for nested JSON / custom endpoints / multimodal input)
• Write correct SQL and PySpark for each function without hallucinating parameter names
  or endpoint names
• Build end-to-end document processing pipelines using the recommended stage order:
  ai_parse_document → ai_classify → ai_extract → ai_query (nested only) →
  ai_similarity
• Use ai_forecast correctly (it's a table-valued function, not a scalar — a common
  mistake)
• Apply failOnError => false and sidecar error tables in batch pipelines

Files Changed

New

File	Contents
databricks-skills/databricks-ai-functions/SKILL.md	Master skill file with YAML front matter, overview, prerequisites, quick start, common patterns, reference index, and common issues table
databricks-skills/databricks-ai-functions/1-task-functions.md	Full reference for all 10 task-specific functions (ai_analyze_sentiment, ai_classify, ai_extract, ai_fix_grammar, ai_gen, ai_mask, ai_similarity, ai_summarize, ai_translate, ai_parse_document) with SQL + PySpark examples
databricks-skills/databricks-ai-functions/2-ai-query.md	Full ai_query reference: when to use it, all parameters with runtime requirements, Foundation Model name table, patterns (structured output, multimodal, SQL UDF, batch error handling)
databricks-skills/databricks-ai-functions/3-ai-forecast.md	Full ai_forecast reference: Pro/Serverless requirement, syntax, output columns, single/multi-group/multi-metric patterns, PySpark via spark.sql()
databricks-skills/databricks-ai-functions/4-document-processing-pipeline.md	End-to-end DLT pipeline example (5 stages), config.yml centralization pattern, near-real-time variant with DSPy + MLflow ChatAgent

Modified

File	Change
databricks-skills/README.md	Added databricks-ai-functions entry to the AI & Agents section
README.md	Updated skill count: 19 → 20