Inspiration
Many developers build incredible tools but struggle to market them effectively. We wanted to bridge the gap between complex source code and high-quality commercial content. We were inspired by the premium aesthetic of Apple product launches and wanted to give every open source contributor the ability to generate that same level of excitement for their own projects automatically.
What it does
RepoStudio transforms a GitHub repository into a professional commercial trailer. It analyzes the actual source code to identify key features and then generates a narrated video. The platform detects the brand colors of the project and uses them to style the entire video. It creates a browser showroom that displays real screenshots of the application alongside animated feature badges that highlight the technical strengths of the repo.
How we built it
The project is built on a stack featuring Next.js for the dashboard and Remotion for the video rendering engine. We integrated Firecrawl and the GitHub API to ingest repository data and source code. ElevenLabs provides the professional narration while Gemini 1.5 Pro acts as the creative director to write scripts based on the code. The entire workflow is managed through Supabase for data persistence and storage.
Challenges we ran into
One major hurdle was ensuring the AI did not hallucinate features that did not exist. We had to move away from simply reading the project description to performing a deep scan of the actual file tree and import statements. Syncing the audio narration with the visual captions also required precise timestamp management to ensure the animations felt snappy and professional rather than drifted.
Accomplishments that we're proud of
We are incredibly proud of the self-referential nature of the project. We successfully used RepoStudio to generate the commercial for RepoStudio itself. The system correctly identified our own tech stack and brand colors to produce a high-fidelity video that looks like it was made by a professional marketing agency.
What we learned
We learned how to translate abstract programming concepts into user-facing value propositions. We gained deep experience in programmatic video editing and learned how to leverage large language models for structured reasoning rather than just text generation. We also discovered the importance of grounding AI creative work in hard data to maintain product authenticity.
What's next for RepoStudio
We plan to introduce social media export formats for platforms like TikTok and LinkedIn. We want to build a GitHub Action that automatically updates the trailer every time a new release is pushed. We also aim to add background music layers and a template gallery so users can choose the specific mood and pacing of their product trailers.
Built With
- claude
- elevenlabs
- firecrawl
- framer-motion
- gemini
- gemini-1.5-pro
- github-api
- nemotron-3
- next.js
- node.js
- playwright
- postgresql
- react
- remotion
- supabase
- tailwind-css
- typescript
- vercel
Log in or sign up for Devpost to join the conversation.