Skip to content

feat: Genie Space quality skill (AI Dev Kit Track 1)#21

Open
stuart-gano wants to merge 2 commits intomainfrom
feat/genie-quality-skill
Open

feat: Genie Space quality skill (AI Dev Kit Track 1)#21
stuart-gano wants to merge 2 commits intomainfrom
feat/genie-quality-skill

Conversation

@stuart-gano
Copy link
Copy Markdown
Collaborator

Summary

  • Adds a coding agent skill for scoring and optimizing Genie Spaces using a 5-stage maturity model (Nascent → Optimized) with 16 criteria
  • Includes standalone example scripts for scoring spaces, adding instructions from UC metadata, and inferring join specs
  • This is Track 1 from the AI Dev Kit integration RFC — makes maturity knowledge available to coding agents without deploying the Workbench app

What's included

File Lines Purpose
skills/.../SKILL.md 158 Main skill — maturity model, criteria tables, assessment workflow, API gotchas
skills/.../optimization.md 321 Stage-by-stage optimization patterns with code examples
skills/.../examples/score-space.py 302 Standalone scorer — fetches space via API, evaluates 16 criteria, prints report
skills/.../examples/add-instructions.py 164 Generates instructions from UC table metadata, applies descriptions
skills/.../examples/add-joins.py 188 Infers join specs from FK naming patterns and shared columns

What it enables

A developer using Claude Code could say:

  • "Score my Genie Space sales_analytics and tell me what to improve"
  • "Add column descriptions to all tables in my space from UC metadata"
  • "Add join specs for my multi-table Genie Space"

Without needing to deploy the Workbench app.

Test plan

  • Verify SKILL.md renders correctly on GitHub
  • Run score-space.py against a real Genie Space
  • Run add-instructions.py against a space with UC tables
  • Run add-joins.py against a multi-table space
  • Verify skill is picked up by Claude Code when working in this repo

Note

The same skill content also exists locally on a branch in the ai-dev-kit repo (feat/genie-quality-skill), ready to push as a PR when that repo accepts contributions.

Three tracks: (1) Genie quality skill for ai-dev-kit, (2) replace
hand-rolled SDK calls with databricks-tools-core, (3) Workbench
MCP server exposing scoring/analysis as MCP tools.
Adds a coding agent skill for scoring, analyzing, and optimizing Genie
Spaces using a 5-stage maturity model (Nascent → Optimized) with 16 criteria.

Includes:
- SKILL.md: maturity model, scoring criteria, assessment workflow, API gotchas
- optimization.md: stage-by-stage fix patterns with code examples
- examples/score-space.py: standalone scorer with formatted report
- examples/add-instructions.py: generate instructions from UC metadata
- examples/add-joins.py: infer join specs from FK naming patterns

This is Track 1 from the AI Dev Kit integration RFC (PR #20). The same
skill content lives in ai-dev-kit for when that repo accepts contributions.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant