Madify

Madify is a local CLI that catalogues photo, PSD, and video files in SQLite, lets you set titles/descriptions/tags, and renames files from those titles. On tag, it also writes an XMP sidecar next to the media file (catalog stays source of truth). The core never touches the filesystem, database, or clock directly — those are injected ports so unit tests stay deterministic.

Features

• Recursive scan into SQLite (images, Photoshop .psd/.psb, common video)
• Tag by id or path — tags merge by default; --replace-tags replaces the set
• XMP sidecar write-back on tag (--no-sidecar to skip)
• List / search the catalog
• Rename from sanitized titles (Clip One → Clip_One.jpg), collisions → _2, _3, …
• Zero runtime dependencies (stdlib only)
• Injectable Clock / FileSystem / CatalogStore / MetadataWriter ports

Install

Requires Python ≥ 3.12 and uv.

git clone https://github.com/sgm-audio/Madify.git
cd Madify
uv sync --group dev
uv run madify

Madify 0.2.1

After PyPI trusted publishing is configured (see below):

uv tool install madify
# or: pip install madify

Quickstart

Keep the catalog outside the media folder so the .sqlite file is not skipped as unsupported media.

# PowerShell demo layout
$demo = Join-Path $env:TEMP "madify-demo"
$dbdir = Join-Path $env:TEMP "madify-db"
New-Item -ItemType Directory -Path $demo, $dbdir -Force | Out-Null
Set-Content "$demo\a.jpg" "x"
Set-Content "$demo\b.psd" "x"
Set-Content "$demo\c.mp4" "x"
Set-Content "$demo\readme.txt" "skip"
$db = Join-Path $dbdir "catalog.sqlite"

uv run madify --db $db scan $demo
uv run madify --db $db tag --id 1 --title "Clip One" --tag demo --tag photo
uv run madify --db $db list
uv run madify --db $db search --query clip
uv run madify --db $db rename --id 1

Commands

Global: --db PATH (default madify.sqlite).

Architecture

Development

uv sync --group dev
uv run pre-commit install
uv build
uv run ruff check .
uv run ruff format --check .
uv run pytest -q --cov=madify --cov-report=term-missing
uv run pdoc madify -d google -o docs/api

API HTML lands in docs/api/ (gitignored).

See CHANGELOG.md, SECURITY.md, and CODE_OF_CONDUCT.md.

PyPI trusted publishing (one-time, manual)

1. Create a PyPI project named madify (or claim the name).
2. On PyPI → Publishing → add a trusted publisher:
   • Owner: sgm-audio
   • Repository: Madify
   • Workflow: publish.yml
   • Environment: pypi
3. On GitHub → Settings → Environments → create pypi (optional protection rules).
4. Publish a GitHub Release — the Publish workflow uploads wheels via OIDC (no API token).

License

MIT — see LICENSE. Copyright (c) 2026 Scott Mills.

Appendix: Supported extensions

Classification is extension-only (case-insensitive). Anything else is skipped on scan.