← Skillful-Alhazen

Curation Skill Builder

Curation Skill Builder

Design and build new TypeDB-backed curation skills using the Skillful Alhazen methodology.

Purpose

The curation-skill-builder skill provides a structured 6-phase framework for designing and implementing new domain skills. It tracks design decisions, schema choices, and evaluation criteria as first-class TypeDB entities — so the design process itself is knowledge-graph-backed.

Use this skill when you want to build a new curation skill from scratch, when you’re modeling a new knowledge domain in TypeDB, or when you want to review and improve an existing skill’s design.

The 6-Phase Framework

Phase Name What you define
1 GOAL System purpose, evaluation criteria, success metrics
2 ENTITY-SCHEMA TypeDB entity, relation, and attribute types for the domain
3 SOURCE-SCHEMA External data sources and how artifacts are captured
4 DERIVATION Ingestion functions (artifact → structured entities)
5 ANALYSIS Query and sensemaking functions (entities → insight)
6 REPORTING Dashboard views and synthesized outputs

Design is iterative — schema gaps discovered during Phase 4 or 5 drive revisions to Phase 2.

Commands

uv run python .claude/skills/curation-skill-builder/skill_builder.py <command> [args]

Domain lifecycle

Command What it does
init-domain Create a domain design tracking project
set-task Set the natural-language task the skill performs
snapshot-skill Snapshot the full skill directory (schema + scripts + prompts + tests)
add-decision Record a design decision and its rationale
export-design Export annotated Markdown design changelog

Phase design

Command What it does
define-goal Define system purpose and evaluation criteria (Phase 1)
add-evaluation Add evaluation criteria to a goal
add-entity-schema Document TypeDB types the domain defines (Phase 2)
add-source-schema Document external data sources and artifact types (Phase 3)
add-derivation-skill Document ingestion functions (Phase 4)
add-analysis-skill Document query and sensemaking functions (Phase 5)
add-phase-spec Attach specs to phase entities
export-design-phases Export full structured phase design report

Gap infrastructure

Command What it does
scaffold-improvement-loop Create GitHub labels, issue templates, and Actions workflows in a skill repo

Typical Workflow

You: I want to build a skill for tracking clinical trial results.
     Help me design it using the 6-phase framework.

You: Initialize a domain called "clinical-trials"
You: Define the goal: track Phase 2/3 trial results for rare disease interventions
You: What entity types do we need in the schema?
You: Add these entity types: clinical-trial, trial-result, trial-cohort
You: What external sources should we query?
You: Export the design phases as a report

Scaffold Gap Infrastructure

For any new skill repository, set up the gap tracking workflow:

uv run python .claude/skills/curation-skill-builder/skill_builder.py \
  scaffold-improvement-loop \
  --repo <owner/repo> \
  --skill <skill-name>

This creates (via GitHub API, no local clone needed):

  • .github/ISSUE_TEMPLATE/skill-gap.md
  • .github/workflows/gap-triage.yml — auto-labels issues on open
  • .github/workflows/weekly-gap-review.yml — Monday gap summary
  • .github/workflows/claude-guidance.yml — posts local fix instructions

See Gap Architecture for the full gap tracking workflow.

Dashboard

The skill contributes a Domain Modeling dashboard at /curation-skill-builder (port 3001 in deployed mode). It shows active domain design projects, phase completion status, and design decision history.

See Also