Claude Code skills for Analytics & Data engineers working with dbt and Snowflake
Altimate Data Skills is a collection of Claude Code skills that encode the workflows and best practices of experienced analytics engineers. These skills transform Claude from a code generator into a capable data engineering assistant.
- 53% accuracy on ADE-bench (43 real-world dbt tasks)
- 3x improvement on model creation tasks vs baseline
- 84% pass rate on Snowflake query optimization (62 TPC-H queries, 1TB dataset)
- 3.6x better performance gains vs baseline (16.8% avg improvement vs 4.7%)
- Skills that teach Claude how to work, not just what to write
/plugin marketplace add AltimateAI/data-engineering-skillsInstall individual skill packs:
# Install dbt skills
/plugin install dbt-skills@data-engineering-skills
# Install Snowflake skills
/plugin install snowflake-skills@data-engineering-skills| Skill | Purpose | Key Behaviors |
|---|---|---|
| creating-dbt-models | Model creation | Convention discovery → Write → Build → Verify output |
| debugging-dbt-errors | Error troubleshooting | Read full error → Check upstream → Apply fix → Rebuild |
| testing-dbt-models | Schema tests | Study existing test patterns → Match project style |
| documenting-dbt-models | Documentation | Analyze model → Generate descriptions |
| migrating-sql-to-dbt | Legacy SQL conversion | Parse SQL → Create proper dbt model |
| refactoring-dbt-models | Safe restructuring | Track dependencies → Apply changes → Verify downstream |
| Skill | Purpose | Key Behaviors |
|---|---|---|
| finding-expensive-queries | Cost analysis | Find and rank queries by cost/time/data scanned |
| optimizing-query-by-id | Performance tuning | Optimize using query ID from history |
| optimizing-query-text | Performance tuning | Profile query → Identify bottlenecks → Apply patterns |
Skills are markdown files that teach Claude how to approach tasks, not just what syntax to use. Each skill has two parts:
When should this skill activate?
---
name: creating-dbt-models
description: |
Guide for creating dbt models. ALWAYS use this skill when:
(1) Creating ANY new model (staging, intermediate, mart)
(2) Task mentions "create", "build", "add" with model/table
(3) Modifying model logic or columns
---What steps should Claude follow?
# dbt Model Development
**Read before you write. Build after you write. Verify your output.**
## Critical Rules
1. ALWAYS run `dbt build` after creating models - compile is NOT enough
2. ALWAYS verify output after build using `dbt show`
3. If build fails 3+ times, stop and reassess your approach
...Skills activate automatically based on your request:
| Your Request | Skill Activated |
|---|---|
| "Create a new orders model" | creating-dbt-models |
| "Fix this compilation error" | debugging-dbt-errors |
| "Add tests to the customers model" | testing-dbt-models |
| "Document the revenue metrics" | documenting-dbt-models |
| "This query is slow, optimize it" | optimizing-query-text |
Skills become even more powerful when combined with Altimate's MCP server. The MCP server provides real-time access to your dbt project and data warehouse:
| MCP Tool | What It Provides |
|---|---|
dbt_project_info |
Project structure, model list, sources |
dbt_model_details |
Column types, dependencies, compiled SQL |
dbt_compile |
Compile models without CLI |
snowflake_query_history |
Recent query executions and stats |
snowflake_table_stats |
Row counts, clustering info |
Evaluated using ADE-bench, a framework for evaluating AI agents on analytics engineering tasks. All tests were run using Claude Sonnet 4.5.
| Configuration | Accuracy | Tasks Resolved |
|---|---|---|
| Baseline Claude (no skills) | 46.5% | 20/43 |
| Claude + Skills | 53.5% | 23/43 |
| Category | Baseline | With Skills | Improvement |
|---|---|---|---|
| Model Creation | 40% | 65% | +25 pts |
| Bug Fixing | 60% | 70% | +10 pts |
| Debugging | 35% | 50% | +15 pts |
| Refactoring | 30% | 35% | +5 pts |
| Analysis | 25% | 30% | +5 pts |
Benchmark on TPC-H 1TB dataset (62 queries) testing optimizing-query-text skill. All tests were run using Claude Sonnet 4.5.
| Configuration | Pass Rate | Avg Performance Improvement |
|---|---|---|
| Baseline Claude (no skills) | 77.4% (48/62) | 4.7% |
| Claude + Skills | 83.9% (52/62) | 16.8% (3.6x better) |
Skills provide structured optimization with query profiling, anti-pattern detection, and semantic preservation validation.
Note: This benchmark uses our internal evaluation framework. We plan to open-source it soon with additional evals.
We welcome contributions! Please see CONTRIBUTING.md for guidelines.
Ideas for contributions:
- New skills for workflows we haven't covered
- Improvements to existing skills based on your team's patterns
- Benchmark results on different datasets
- Bug reports and feature requests
We're actively developing:
- Airflow skills — DAG development, debugging, testing
- Cross-platform migration — dbt to/from SQL Server, Oracle
- Snowflake cost optimization — Warehouse sizing, query patterns
- Data quality workflows — Anomaly detection, freshness checks
This project is licensed under the MIT License - see the LICENSE file for details.
Built by the team at Altimate AI — Making data engineering delightful.