A Claude skill that grounds Claude in accurate knowledge for writing and optimising SQL queries against AWS CUR 2.0 data enriched with SPRUCE greenops columns.
When loaded, the skill makes Claude reliably accurate on:
- CUR 2.0 schema details (column naming, nested arrays, conditional columns)
BILLING_PERIODpartitioning semantics- Engine-specific syntax for Athena and DuckDB
- All SPRUCE column meanings and units (carbon, electricity, water)
- Download the latest
spruce-sql-advisor.skillfrom the Releases page. - In Claude.ai, go to Settings → Customize → Skills, click on +, Create skills and Upload a skill.
- Start a new chat and ask a CUR 2.0 SQL question — Claude will load the skill
automatically. Alternatively, call it explicitely by typing
/spruce-sql-advisorin a new chat window.
Full documentation, including example prompts and screenshots, is at digitalpebble.github.io/spruce-sql-advisor.
spruce-sql-advisor/ ← repo root
├── spruce-sql-advisor/ ← the skill itself (gets packaged into .skill)
│ ├── SKILL.md
│ ├── LICENSE, NOTICE
│ └── references/
├── docs/ ← GitHub Pages site
├── .github/workflows/ ← CI: validate + release
├── README.md, LICENSE, NOTICE
The skill folder name is repeated by design: package_skill.py uses the folder name
both as the artefact name (spruce-sql-advisor.skill) and as the top-level path inside
the zip, and that name must match the skill's name frontmatter field.
git clone https://github.com/DigitalPebble/spruce-sql-advisor
cd spruce-sql-advisor
# Fetch Anthropic's packaging script
git clone --depth 1 --filter=blob:none --sparse \
https://github.com/anthropics/skills _anthropic_skills
git -C _anthropic_skills sparse-checkout set skills/skill-creator/scripts
PYTHONPATH=_anthropic_skills/skills/skill-creator \
python3 -m scripts.package_skill ./spruce-sql-advisor .Apache License 2.0 — see LICENSE and NOTICE.
- SPRUCE — the open-source GreenOps platform this skill is designed to work with.
- DigitalPebble — green software consultancy.