Git Story Film
Use repository history as the evidence for a film
Choose a story from real commits, then explain the project’s growth through characters, storyboards and hand-drawn animation.

Entry points
- Claude Code
- Coding assistants that can read the complete Skill and work on local files
- Repository stars
- 5,916
- Checked
- Categories
- Design & VideoEngineering
Start by approving the direction
Choose a repository with a few clear milestones and decide on 3–5 changes the audience should remember. Settle the evidence and narrative before character work, while it is still easy to correct an appealing story that misrepresents the product.
Pick a task and get started
These four tasks group the seven production stages. Install the Skill and start with the first task, then continue with the approved artifacts. Replace the angle-bracket placeholder with the real repository URL.
Start with repository evidence and stop at a reviewable direction.
Copy the complete Skill into a separate workspace
An editorial shell sequence based on the author’s complete-folder installation, pinned to the reviewed commit. It needs Git and a macOS/Linux/WSL shell. Use a fresh location and inspect existing names first. It sparsely checks out the Skill, then copies it and the root notices into the workspace without installing Raven. Native Windows users can copy the complete folder as documented; production dependencies come later.
Read the author’s setup guidegit clone --filter=blob:none --sparse https://github.com/EverMind-AI/Raven.git raven-skill-source git -C raven-skill-source sparse-checkout set skills/git-story-film git -C raven-skill-source checkout --detach 9d9bec90bcd76186d2981fd23b3b6280d83e3684 mkdir -p film-workspace/.claude/skills cp -R raven-skill-source/skills/git-story-film film-workspace/.claude/skills/ cp raven-skill-source/LICENSE raven-skill-source/NOTICES.md film-workspace/.claude/skills/git-story-film/ cd film-workspaceBegin in film-workspace
Open film-workspace in Claude Code, reload the Skill and confirm that .claude/skills/git-story-film/SKILL.md is readable. Check Git and Python first. Replace the repository URL in the first prompt and state the branch to cover. Keep the target checkout outside this workspace, for example at ../target-repo. From the workspace, the statistics command is python3 .claude/skills/git-story-film/scripts/git_timeline.py ../target-repo --branch HEAD --out timeline.json; HEAD should be your selected target revision. The first round needs story research only. Prepare rendering tools before character production.
Author quick start and seven stagesGive your AI this first task
These are editorial tasks based on the pinned source. Results describe expected deliverables and checks. We did not run these commands, draw keyframes or render a film.
/git-story-film <repository URL> Plan a roughly two-minute English introduction for people new to the product, in 16:9. Keep the target checkout outside the current film-workspace and use the selected branch’s full history, not a shallow clone. Record the branch and full HEAD hash. Run scripts/git_timeline.py from the complete Skill, writing timeline.json in the current film workspace, then check candidate events against the target README, feature documentation and public release records. Leave the target repository unchanged. Deliver timeline.json and a first-stage OUTLINE.md containing a logline, a suitable story form, protagonist and caption voice, four-part arc, 3–5 takeaways, and a mapping from facts to visible actions. Source every figure, release and turning point; do not invent a revert if there is none. List unresolved audience, logo or factual questions. Wait for direction approval before character work; do not install rendering dependencies or build the film in this round.
Check what you receive
- timeline.json and OUTLINE.md in the film workspace, explaining who the product helps and why the selected events matter.
- Statistics identify the branch, HEAD and date basis. Key features, public release dates and reverts each have traceable evidence; unresolved claims remain visible.
- Approve the direction, language, aspect ratio and logo treatment before character work. This first task delivers a plan.
More about this skill
Git Story Film is for launches, anniversaries and introductions to a software project. An assistant selects meaningful commits, releases and reversals, translates features into visible actions, and shapes a film of roughly two minutes. The story explains what the character is trying to do, what changes along the way, and what the product can now do for its users.
It is a separate Skill directory inside Raven, with a Canvas drawing engine, an HTML review Studio, Git statistics helpers and local rendering scripts. The coding assistant researches the project and adapts the film code; a browser draws frames and FFmpeg encodes them. This workflow does not require installing the Raven application or connecting a text-to-video service.
The author defines seven approval stages: direction, cast, full outline, the first 30 seconds of storyboards, the complete build, Studio review and final delivery. Captions provide the narrative voice over a Web Audio score. Establish the opening’s tone and timing before extending the film and producing the high-resolution export.
Sources and verification
Sources checked
Pinned to commit 9d9bec9. Read 32 author documents and key source files in full, with an additional targeted review of the engine runtime. Documentation and static checks only; no installation, playback or rendering.