backlog-atlas 0.17

Generate a self-hosted backlog snapshot and dashboard for maintainers.

Backlog Atlas

Backlog Atlas generates a self-hosted backlog snapshot and static dashboard for GitHub repository maintainers.

It keeps GitHub Issues as the source of truth, derives backlog state from open issues and linked pull requests, and publishes the result to a dedicated backlog-atlas branch. The generated branch is machine-owned and can be served directly with GitHub Pages.

Documentation

• User Guide — install Backlog Atlas on a repository, enable the web UI, run updates, and uninstall.
• Maintainer Guide — local development, testing, packaging, and release prep for this package.
• Design Note — how snapshots, diffs, and the update log fit together.
• Changelog — user-facing release notes.

What It Produces

The backlog-atlas branch contains:

• backlog.json — structured backlog data plus the recent activity tail used by the web UI.
• updates.jsonl — append-only structured activity history.
• last_snapshot.json — internal diff state for the next run.
• index.html — the static dashboard.
• favicon.svg — the dashboard icon.
• badge.svg — a README badge image that can link to the hosted dashboard.

The dashboard can also load a generated atlas.json manifest next to index.html and federate multiple public backlog.json files in the browser. Author the multi-repo source as YAML, then compile it to the browser manifest.

The default branch only needs the installed workflow, manifest, and config:

• .github/workflows/update-backlog-atlas.yml
• .github/backlog-atlas/manifest.json
• .github/backlog-atlas/config.yaml — editable repository configuration

Quick Start

pip install backlog-atlas
cd /path/to/target-repo
backlog-atlas install --dry-run
backlog-atlas install

The dry run previews the files and install source before anything is written. Install is safe to rerun: it first removes old Backlog Atlas hooks/manifests and preserves config and generated dashboard history. When the previous install manifest lists bundled wheels, it writes a temporary cleanup workflow that removes those old wheels only after the new install lands. The manifest is the cleanup source of truth: it records installed files and which ones are removed only by clean uninstall. Review and push the install commit that Backlog Atlas creates, then enable GitHub Pages from the backlog-atlas branch. See the User Guide for the full local and remote install flows.

After the dashboard is hosted, add a Backlog Atlas badge to the repository README by wrapping the generated badge image in a link to the dashboard. Assuming the generated badge.svg is hosted alongside the dashboard, replace DASHBOARD_URL with the dashboard URL:

[![Backlog Atlas](DASHBOARD_URL/badge.svg)](DASHBOARD_URL)

For a conventional GitHub Pages dashboard, that is usually:

[![Backlog Atlas](https://OWNER.github.io/REPO/badge.svg)](https://OWNER.github.io/REPO/)

CLI

# install the workflow and manifest in a repository
backlog-atlas install [flags]

# remove installed hooks/manifests; --clean also removes config and branch
backlog-atlas uninstall [flags]

# write the static web UI files for preview or packaging
backlog-atlas dump-web --output PATH

# compile YAML multi-repo config into atlas.json for the browser UI
backlog-atlas dump-atlas --config atlas.yaml --output PATH

# manage repos tracked by .github/backlog-atlas/atlas.yaml
backlog-atlas atlas list
backlog-atlas atlas add owner/name [--backlog-url URL]
backlog-atlas atlas remove owner/name
backlog-atlas atlas install --delivery pr|push
backlog-atlas atlas install --local --checkout-root PATH

# refresh backlog data; normally run by the installed GitHub Action
backlog-atlas update [flags]

# explain how one issue is classified by the active config
backlog-atlas classify ISSUE_NUMBER [--repo URL]

Development Checks

Maintainers can install local check tools with:

python -m pip install -e ".[dev]"
python -m black --check backlog_atlas tests tools
python -m pyflakes backlog_atlas tests tools
python -m mypy
python -m pytest

See the Maintainer Guide for the full development workflow.

License

MIT.