pr-to-video

Converts GitHub pull requests into narrated code-change explainer videos with voiceover, captions, and BGM.

Updated Jun 20, 2026
One-click install
npx skills add https://github.com/Noetfield-Systems/SourceA --skill pr-to-video-noetfield-systems
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: pr-to-video
Source: https://github.com/Noetfield-Systems/SourceA/tree/main/commercial-video-factory/trustfield-storyboard-hf-v1/.agents/skills/pr-to-video
Command: npx skills add https://github.com/Noetfield-Systems/SourceA --skill pr-to-video-noetfield-systems

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires puppeteer-core, google-genai, transformers, torch, soundfile, numpy, elevenlabs, and includes scripts (resource) and assets (resource) components.

What problem does it solve? Turning a merged pull request into a clear, shareable video explanation normally requires manual scripting, recording, and editing. This Skill automates the entire pipeline: it reads a GitHub PR via the gh CLI, designs a story from the diff, generates narration audio and background music, builds animated scenes, and renders a finished MP4. ## Core Features & Use Cases - PR Ingest via gh CLI: Pulls PR title, body, diff, commits, files, and +/- stats, plus contributor avatars for an optional credits scene — no website scraping involved. - Deterministic Pipeline Stages: Story design, audio synthesis (HeyGen, ElevenLabs, or local Kokoro TTS), caption grouping, scene composition, and final assembly are driven by scripts and subagents with validation gates at each step. - Styled Output: Uses the claude design preset with code-window, before-after, and diff-highlight scenes; supports landscape, portrait, and square aspect ratios. - Use Case: A developer advocate wants a 60-second changelog video for PR #1187. They provide the PR URL, confirm the brief, and the pipeline produces a narrated explainer video showing what shipped and why. ## Quick Start Ask the assistant to turn the GitHub pull request at a given URL into a short explainer video, then answer the one-round brief about angle, audience, and length.

Frequently Asked Questions about pr-to-video

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

FAQPage Schema
How do I turn a GitHub pull request into a video?

Provide a PR URL, an owner/repo#N reference, or say "this PR" inside a checked-out repo. The pipeline reads the PR with the gh CLI, designs a story from the diff, generates voiceover and background music, builds scenes, and renders an MP4.

What inputs does the pr-to-video workflow accept?

It accepts only GitHub pull requests: a full PR URL, an <owner>/<repo>#<N> reference, or "this PR" in a checked-out repository. Marketing sites, general websites, and plain text topics are routed to other skills instead.

Does PR video generation work without API keys?

Yes. Without HeyGen or ElevenLabs keys it falls back to local Kokoro TTS, and without a Gemini key it generates BGM locally with MusicGen. The gh CLI must still be authenticated to read the PR.

Can it make videos from private GitHub repositories?

Yes, as long as gh auth status passes with an account that has access to the private repository. If gh returns an auth or not-found error, the pipeline stops and reports the exact stderr rather than fabricating PR contents.

What video formats and lengths does it support?

It targets up to roughly 3 minutes with a sweet spot of 30-90 seconds, and supports 16:9 landscape (1920x1080), 9:16 portrait (1080x1920), and 1:1 square (1080x1080) aspect ratios. Longer exhaustive walkthroughs are routed to a different skill.

Why did the rendered video come out silent?

A silent render means voice wavs were not wired into the assembly, usually because prep was run without the audio_meta.json flag. Re-run prep with --audio-meta ./audio_meta.json and reassemble; the scripts print a CRITICAL banner when this wiring breaks.