Skip to content

As a nabledge skill developer, I want comprehensive mapping from official docs to knowledge files so that AI agents can automatically create categorized knowledge files #10

@kiyotis

Description

@kiyotis

Situation

Nablarch official documentation exists across multiple repositories (nablarch-document, nablarch-system-development-guide) with hundreds of files. We need to create knowledge files for the nabledge-6 skill, but there is no mapping information defining which official documentation files map to which knowledge files and what category each file belongs to.

Pain

Skill developers must manually identify and classify documentation files. AI agents cannot automatically create knowledge files without clear mapping. Users cannot filter documentation by specific categories (e.g., "process only nablarch-batch and handlers files"). There is no systematic way to trace generated knowledge back to official sources.

Benefit

  • Skill developers can filter documentation by specific category IDs for targeted knowledge file creation
  • AI agents can automatically generate knowledge files using clear category-to-file mappings
  • Users can trace knowledge back to official Japanese documentation via URLs
  • Knowledge file creation workflow becomes efficient by separating asset collection from content conversion

Success Criteria

  • Skill developers can filter documentation by specific category IDs (e.g., nablarch-batch, handlers, restful-web-service) to select only relevant files for processing
  • AI agents can read mapping files to automatically identify which official documentation files map to which knowledge files for each category
  • Users can navigate from any generated knowledge file to its official Japanese documentation source using provided URLs
  • Mapping files cover all expected documentation files (v6: 337 files, v5: 431 files)
  • Mapping accuracy is verifiable through automated validation (file counts, category definitions, paths, URLs)
  • Excel export enables non-technical stakeholders to review and verify mappings efficiently

Design Reference: See doc/mapping-file-design.md for detailed specifications on:

  • File structure (categories JSON, mapping JSON)
  • Classification taxonomy (21 category IDs across 6 types)
  • Official URL conversion rules
  • Source file scope and exclusions
  • Asset reference directives
  • Excel export format
  • Knowledge file creation considerations

Metadata

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions