databricks-ai-functions
Use Databricks built-in AI Functions (ai_classify, ai_extract, ai_summarize, ai_mask, ai_translate, ai_fix_grammar, ai_gen, ai_analyze_sentiment, ai_similarity, ai_parse_document, ai_query, ai_forecast) to add AI capabilities directly to SQL and PySpark pipelines without managing model endpoints. Also covers document parsing and building custom RAG pipelines (parse → chunk → index → query).
88
85%
Does it follow best practices?
Impact
Pending
No eval scenarios have been run
Passed
No known issues
Quality
Discovery
82%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a strong description with excellent specificity, listing all 12 Databricks AI functions by name and covering concrete use cases including RAG pipeline construction. The main weakness is the absence of an explicit 'Use when...' clause, which means Claude must infer when to select this skill rather than having clear trigger guidance. The domain-specific terminology makes it highly distinctive and unlikely to conflict with other skills.
Suggestions
Add an explicit 'Use when...' clause, e.g., 'Use when the user asks about Databricks AI Functions, adding AI capabilities to SQL or PySpark queries, building RAG pipelines in Databricks, or references any of the ai_* built-in functions.'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description lists 12 specific AI functions by name (ai_classify, ai_extract, ai_summarize, etc.) and describes concrete use cases like adding AI capabilities to SQL/PySpark pipelines, document parsing, and building custom RAG pipelines with a clear workflow (parse → chunk → index → query). | 3 / 3 |
Completeness | The 'what' is thoroughly covered with specific functions and use cases, but there is no explicit 'Use when...' clause or equivalent trigger guidance telling Claude when to select this skill. The when is only implied by the capabilities listed. | 2 / 3 |
Trigger Term Quality | Includes highly natural and specific trigger terms users would actually say: 'Databricks', 'AI Functions', each function name individually, 'SQL', 'PySpark', 'RAG', 'document parsing', 'model endpoints'. These cover a wide range of how users might phrase requests related to this skill. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive due to the specific Databricks platform context, named AI functions, and the SQL/PySpark pipeline focus. This is unlikely to conflict with generic AI, SQL, or document processing skills because of the very specific Databricks built-in function names. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
87%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a high-quality skill that efficiently covers a broad set of Databricks AI Functions with concrete, executable examples and excellent progressive disclosure. The function selection decision tables are particularly valuable, providing clear guidance on when to use each function. The main weakness is the lack of explicit validation checkpoints within multi-step workflows, especially for batch write operations.
Suggestions
Add a validation/verification step to Pattern 2 (PII Redaction) and Pattern 3 (Document Ingestion) — e.g., sample a few rows to confirm masking worked before writing to the target table, or check parse_error counts before proceeding.
Consider adding a brief 'Batch job workflow' section with explicit checkpoints: test on sample → validate outputs → run at scale → verify row counts, to address the missing feedback loops for batch operations.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is lean and efficient. The overview paragraph is slightly explanatory but earns its place by establishing the key mental model (task-specific vs general-purpose vs table-valued). No unnecessary explanations of what SQL or PySpark are, no padding about AI concepts Claude already knows. The tables are dense and information-rich. | 3 / 3 |
Actionability | Every pattern includes fully executable SQL and/or PySpark code that is copy-paste ready. The quick start shows a complete multi-function query, and patterns cover real scenarios (PII redaction, document ingestion, semantic matching, forecasting) with concrete, runnable examples including schema definitions and write operations. | 3 / 3 |
Workflow Clarity | The patterns are clearly sequenced (e.g., Pattern 3 shows read → parse → filter errors → enrich), and the function selection tables provide clear decision logic. However, there are no explicit validation checkpoints or error recovery feedback loops for batch operations — the troubleshooting table is reactive rather than integrated into workflows. For batch/destructive operations like Pattern 2 (writing redacted data to Delta), there's no verification step. | 2 / 3 |
Progressive Disclosure | Excellent structure: the SKILL.md provides a concise overview with quick start and common patterns, then clearly signals four one-level-deep reference files covering task functions, ai_query, ai_forecast, and a full document processing pipeline. Each reference file link includes a descriptive summary of its contents. | 3 / 3 |
Total | 11 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- databricks-solutions/ai-dev-kit
- Commit
02aac8c
- Reviewed
Table of Contents
Is this your skill?
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.