Skip to content

CLAUDE.md

Latest commit

 

History

History
63 lines (48 loc) · 2.34 KB

CLAUDE.md

File metadata and controls

63 lines (48 loc) · 2.34 KB
Raw

Claude Code Project Configuration

Project Overview

This is the OGD to LOD project - a tool to create RML (RDF Mapping Language) mappings for CSV files using generative AI.

Allowed Operations

Git Operations

  • git add, git commit, git push - for committing and pushing changes
  • git rm, git reset - for managing staged files

GitHub CLI

  • gh issue create - create issues in the repository
  • gh label create - create labels for issues

Python Development

  • python3 - run Python scripts
  • pip install - install dependencies
  • pytest - run tests
  • ogd-to-lod - run the CLI tool

Environment

  • GitHub token available via $APP_GITHUB_TOKEN environment variable (from .env) — this is for the application runtime only
  • When acting as Claude Code agent (e.g. using gh CLI), use $GITHUB_TOKEN from the shell environment directly — never source .env or use APP_GITHUB_TOKEN for agent actions
  • Azure OpenAI credentials via $AZURE_OPENAI_ENDPOINT and $AZURE_OPENAI_KEY

Project Structure

ogd-to-lod/
├── src/ogd_to_lod/    # Main package
│   ├── parsers/       # CSV and DCAT parsers
│   ├── ai/            # Azure OpenAI integration
│   ├── graph/         # LangGraph conversation flow
│   ├── rml/           # RML generation (prompts, generator)
│   ├── github/        # GitHub PR creation
│   └── validation/    # Two-tier RML validation (syntax + RMLMapper)
├── tests/             # Test files
├── config/            # Configuration files
├── scripts/           # Utility scripts (RMLMapper setup, worktrees)
├── brainstorm.md      # Project requirements
└── architecture.md    # Technical architecture

Key Documentation

  • brainstorm.md - Project goals, milestones (MVP, V1, Later), and scope
  • architecture.md - Tech stack, conversation flow, LangGraph states, prompting strategy

GitHub Repository

  • Repository: redlink-gmbh/ogd-to-lod
  • Issues: MVP issues #1-#10 are created

Development Workflow

  1. Create a feature branch from main (e.g., feature/issue-37)
  2. Work on issues sequentially — one issue at a time in the main repo directory
  3. Run tests with pytest
  4. Commit with descriptive messages referencing issue numbers
  5. Push branch and create a Pull Request to main
  6. Never commit directly to main branch