bilder

Regenerates bilingual manual screenshots of the application window before each release.

Updated Aug 22, 2026
One-click install
npx skills add https://github.com/Bascht74/videopodcast-magic --skill bilder-bascht74
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: bilder
Source: https://github.com/Bascht74/videopodcast-magic/tree/main/.claude/skills/bilder
Command: npx skills add https://github.com/Bascht74/videopodcast-magic --skill bilder-bascht74

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires PySide6.

What problem does it solve? The pictures in the manual drift out of date whenever the application window changes, and retaking them by hand is slow, inconsistent, and easy to forget before a release. This Skill defines the complete screenshot pass: when to run it, what it needs, how to shoot, and how to verify the results. ## Core Features & Use Cases - Repeatable automated screenshots: Uses docs/notes/shoot_screenshots.py to load the program as a module, open a project built from test fixtures, and save window.grab() output in English and German (docs/images/NAME.png and NAME.de.png). - Release-gated workflow: Pictures are taken once per version, not after every change; outdated pictures are noted in between and refreshed in one pass before release. - Quality checklist: Eight verification questions catch cut-off labels, wrong-language catalogue entries, unparsable assembled sentences, leaked keys or real names, and unexpected changes. - Use Case: Before shipping a release, run the caption test with VPM_LAYOUT_PLATFORM=cocoa, build the fixtures, shoot the main, split, and channels sets in both languages, compare checksums, copy only changed pictures, and record what invalidated each one. ## Quick Start Run the pre-release screenshot pass for the manual: build the test fixtures, shoot all picture sets in English and German, compare checksums, and copy only the images that actually changed into docs/images.

Frequently Asked Questions about bilder

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I regenerate the manual screenshots for a release?▼

Build the fixtures with `cd tests && bash fixtures.sh`, then run `shoot_screenshots.py` for each set (main, split, channels) in both languages via `VPM_SHOT_SET` and `LANGUAGE`. Compare checksums and copy only changed pictures into `docs/images/`.

How are automated Qt window screenshots taken without a desktop?▼

The script loads the program as a module, opens a project built from test fixtures, and saves `window.grab()` output. The window uses `WA_DontShowOnScreen`, so nothing appears on the desktop, though the application still activates its space on macOS.

Can the screenshot pass run over SSH or without a logged-in screen?▼

No. The cocoa platform talks to the window server, so a logged-in graphical session on the Mac is required; the script stops otherwise. `VPM_SHOT_PLATFORM=offscreen` exists as an emergency exit, but its Fusion-look pictures must not go into the manual.

Why run the window_captions_fit test before taking screenshots?▼

Offscreen rendering draws Fusion font faces, so only a cocoa run shows whether the system's real typewriter face fits each caption column. A red line there is fixed in the program before any picture is taken.

What must never appear on a manual screenshot?▼

No real production names, presets, production IDs, private disk paths, people's names, or keys may appear. Only fixture names and `/tmp` paths are allowed, and the script nets `load_api_key`, `AUPHONIC_TOKEN`, and network-reaching functions to enforce this.

When should outdated manual pictures be retaken?▼

Not after every window change, but once before each release. In between, an outdated picture is only noted; the full pass then reshoots everything, updates descriptions, and records which change invalidated each picture.