How sounddrill31/crave_aosp_builder works
Project type and purpose
• This is a CI/CD-style tooling repository (“crave_aosp_builder”) built around GitHub Actions.
• Its purpose is to orchestrate Android OS builds on external build servers (Crave.io) or on your own self-hosted runners, with optional integrations (code-server remote IDE, Telegram notifications, GitHub Releases, Backblaze B2 signing).
File structure
• Top-level:
• .github/workflows/ – multiple workflow YAMLs: configure self-hosted runner, start/restart it, and the main build pipelines.
• configs/ – templates for code-server and Crave CLI (crave.yaml.*).
• scripts/ – bash scripts for environment checks, runner setup, code-server launch, VSCode tunneling.
• assets/ – documentation images (secrets setup, runner token flow).
• README.md & sample crave.conf.sample – detailed usage and config reference.
• No “frontend” or “backend” per se; instead it’s an “automation orchestration” project.
README for architecture clues
• Describes workflows dispatch, required secrets (CRAVE_USERNAME, CRAVE_TOKEN, optional TELEGRAM_TOKEN, etc.).
• Two modes: GitHub-hosted runners (6h timeout, more Action abuse risk) vs self-hosted runners (persist beyond timeouts).
• External dependencies: Crave.io build servers, optional Backblaze B2, Telegram Bot API, GitHub Releases API.
Elements
• Labels: e.g., “GitHub Repository”, “GitHub Actions”, “Self-Hosted Runner”, “Crave Build Service”, “Artifact Storage”
• from Repo → Actions → Runner
• from Runner ↔ Crave.io
• from Runner → GitHub Releases / Telegram API / Backblaze B2
• Orange clouds for external services (Crave.io, Telegram, Backblaze)
Final advice
• Be very detailed: break the runner node into sub-components (setup scripts, code-server, env checker).
• Optional paths (e.g., signing, custom YAML override).
• Secrets as configuration inputs.