Skip to content

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

GitHub Automation and CI — DermaLens AI

This directory contains the GitHub Actions automation for DermaLens AI. It is not the project's main README.

Start here: Project overview · Source and architecture · Persistent development tracker · Deployment instructions

Why this directory exists

DermaLens has a Next.js frontend, a FastAPI backend, research utilities and optional integrations. CI checks that code changes do not break shared contracts, type checks, tests or container packaging before they reach the default branch.

These checks validate software engineering, not medical accuracy. A successful workflow does not imply that the six-class disease classifier is connected, externally validated or deployed as a medical service.

What runs automatically?

Workflow Trigger What it checks
Phase 0 — Architecture contracts Pull requests; pushes to main and the original Phase 0 branch Python lint, required architecture documents, model manifest schema, OpenAPI contract tests
Phase 1 — Backend and frontend Pull requests; pushes to main and the original Phase 1 branch Python setup, Ruff, formatting, mypy, pytest, repository hygiene; web npm install, ESLint, Vitest, Next.js production build and TypeScript
Phase 8 — Container configuration Pull requests; pushes to main Valid Compose configuration and exclusions for credentials, face datasets and model weights
Phase 12 — Disposable Docker staging smoke Pull requests; manual workflow_dispatch Build/run local Docker services, apply real database migrations, check API liveness, intentional fail-closed model readiness, and web response; clean up disposable CI containers

The exact workflow definitions—not this summary—are the source of truth for job names, triggers and steps.

CI workflow at a glance

flowchart TD
    A[Feature branch] --> B[Open pull request]
    B --> C[Architecture and API contracts]
    B --> D[Python backend checks]
    B --> E[Next.js frontend checks]
    B --> F[Compose config and Docker smoke]
    C --> G{All required checks pass?}
    D --> G
    E --> G
    F --> G
    G -->|No| H[Inspect logs and fix on branch]
    H --> B
    G -->|Yes| I[Review PR and merge]
    I --> J[Run push-triggered main checks]
Loading

No workflow here publishes a medical inference model or automatically provides a public Vercel URL. Deployment is a separate, explicitly configured process.

How to contribute

  1. Read the root README, GEMINI.md, and relevant architecture documentation.
  2. Create a feature branch from the latest main. Preserve any existing uncommitted changes.
  3. Make a focused change, add or update meaningful tests, and run the corresponding local commands below.
  4. Open a pull request and check the Actions page for that PR's exact commit.
  5. Fix failing checks. Record what passed, what is blocked, and what was not tested. Merge only after review and the required green CI results.
  6. Update GEMINI.md with verified progress, remaining work and the next task when the change affects project status.

See the child workflows reference for common failure diagnosis.

Run checks locally

From the repository root with Python 3.11 and uv:

uv sync --python 3.11 --frozen --all-groups
uv pip install --python .venv/bin/python -r requirements-persistence.txt -r requirements-cv.txt -r requirements-rag.txt
uv run --frozen --no-sync ruff check .
uv run --frozen --no-sync ruff format --check apps packages services tests/test_api_foundation.py tests/test_phase2_backend.py
uv run --frozen --no-sync mypy
uv run --frozen --no-sync python -m scripts.validate_phase0
uv run --frozen --no-sync python -m pytest -q
uv run --frozen --no-sync python scripts/check_repository_hygiene.py

From apps/web/ with Node.js 22+:

npm ci
npm run lint
npm test
npm run build
npm run typecheck

To validate a disposable local Docker stack, see infra/docker/README.md and the staging workflow. Do not use docker compose down --volumes on a development database whose contents you need.

Security and release boundaries

  • Never commit .env files, passwords, authentication tokens, encryption keys, patient photographs, private datasets, or trained model checkpoints.
  • Keep the original research training dependencies separate from the FastAPI and Next.js production images.
  • Check the exact SHA of tested code before claiming a green build. CI output may contain warnings that need follow-up.
  • Public model inference remains disabled. The classifier needs real authorized weights, exact label ordering, parity verification and independent evaluation before any research activation, and clinical deployment requires separate safety/legal review.
  • Vercel web deployment and optional container API deployment are independent; neither happens just because this CI folder exists.

Current verified context

On October 25, 2026, engineering PR #37 passed 133 Python tests and 17 frontend tests, plus build/static checks and Docker smoke. Later documentation PRs #38 and #39 were merged following passing checks. Recheck the current Actions runs before relying on this dated snapshot.

For the latest task-level progress, use GEMINI.md—not this CI documentation.

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages