Skip to content

Polish open-source setup and documentation#11

Open
Spbd1 wants to merge 1 commit into
codex/build-review,-evaluation,-and-report-workflowsfrom
codex/polish-project-for-github-open-source-readiness
Open

Polish open-source setup and documentation#11
Spbd1 wants to merge 1 commit into
codex/build-review,-evaluation,-and-report-workflowsfrom
codex/polish-project-for-github-open-source-readiness

Conversation

@Spbd1
Copy link
Copy Markdown
Owner

@Spbd1 Spbd1 commented May 18, 2026

Motivation

  • Make the repository GitHub-ready by improving install/run/test ergonomics, adding practical docs, and providing a one-command developer experience.
  • Provide a safe, local-first demo workflow with seeded demo inputs and a small benchmark so contributors can iterate on taxonomy and rules without a DB.
  • Add containerization and environment guidance so users can run the MVP reproducibly in local or containerized environments.

Description

  • Expanded the README and added practical documentation files (docs/*.md) that cover architecture, taxonomy design, taxonomy expansion protocol, annotation guidelines, evaluation protocol, API contract, dashboard user guide, limitations, and roadmap.
  • Added container and runtime artifacts including Dockerfile, docker-compose.yml, and .env.example, plus a CHANGELOG.md and packaged startup improvements to keep the MVP database-free.
  • Improved scripts and dev flow: enhanced scripts/dev.py to install deps, seed demo data, optionally import a taxonomy workbook, run backend and frontend, and open the browser; made scripts/run_backend.py env-configurable; added scripts/import_taxonomy_excel.py and scripts/export_taxonomy_excel.py fixes; and improved frontend dev server to accept host/port from args/env.
  • Seeded practical demo and benchmark data (data/examples/demo_inputs.jsonl and data/benchmarks/mini_eval_set.jsonl), updated scripts/seed_demo_data.py to write artifacts only when missing, and applied small engine/pack fixes (import formatting, exporter attribute assignment, pack manager/importer tidy, and a reusable uvicorn stub for local startup/testing).

Testing

  • Ran the test suite with make test (pytest): all tests passed (42 passed, 4 warnings).
  • Ran lint/format checks with ruff: all checks passed after automatic fixes.
  • Ran the evaluation harness with make evaluate / python scripts/run_evaluation.py: benchmark executed and produced metrics (precision/recall/F1 and error buckets) to serve as regression signals.
  • Built the frontend with npm --prefix frontend run build and executed python scripts/dev.py --install to verify dependency installation and seeding; both commands completed successfully in this environment, while docker compose could not be exercised here because Docker is not available.

Codex Task

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant