CapYap

Local-first YouTube transcript assistant with citation-grounded answers.

• Paste a YouTube link (or transcript file)
• Ask questions with AI-powered, timestamp-cited answers
• Click timestamps to jump to the exact moment in the video
• Use native or AI-generated chapters
• Export transcripts as TXT, JSON, HTML, or PDF

Privacy: Your API key is used per-session and never stored to disk. Everything runs locally.

---

Download

Desktop App

Platform	Download	Format
macOS	Capyap_macOS.app.tar.gz	.app bundle
Windows	CapYap_1.0.4_x64-setup.exe	Installer (.exe)
Windows	CapYap_1.0.4_x64_en-US.msi	Installer (.msi)
Linux	CapYap_1.0.4_amd64.deb	Debian/Ubuntu
Linux	CapYap_1.0.4_amd64.AppImage	AppImage (portable)
Linux	CapYap-1.0.4-1.x86_64.rpm	Fedora/RHEL

All installers: GitHub Releases

The desktop app auto-starts the local backend at 127.0.0.1:8000 on launch. If auto-start fails, run manually: conda activate capyap && capyap start --no-browser

Web App (localhost)

Run CapYap in your browser without the desktop shell:

git clone https://github.com/arashabadi/capyap.git
cd capyap
conda env create -f capyap.yml
conda activate capyap
npm run build
capyap start

This installs frontend deps (if needed), builds static assets, launches the local backend + UI, and opens your browser.

---

Free Local Agents (Ollama)

Run CapYap fully local with Ollama — no cloud API key required. Setup guide: docs/OLLAMA_SETUP.md

---

First Run

1. Paste a YouTube URL or local transcript .txt path.
2. Click Start AI Analysis.
3. Choose a provider and paste your API key (or choose ollama for free local inference).
4. Ask questions and review timestamp citations.

Daily Usage

conda activate capyap
capyap start

Flags: --no-browser | --port 8080 | --host 0.0.0.0

---

Conda Environments

File	Env Name	Purpose
capyap.yml	capyap	Production — run the app (Python + Node)
capyap-dev.yml	capyap_dev	Development — adds Rust, ffmpeg, pytest
apps/environment.yml	capyap_apps	Full app stack — backend + frontend + desktop build chain

# Production
conda env create -f capyap.yml

# Development (includes build toolchain)
conda env create -f capyap-dev.yml

For desktop dev, also install frontend/desktop deps:

conda activate capyap_dev
npm --prefix apps/frontend install
npm --prefix apps/desktop install --include=dev

---

Export Options

Format	Description
Clean Text (.txt)	Plain transcript text
With Timestamps (.txt)	Transcript with [mm:ss] markers
Raw JSON (.json)	Structured transcript, metadata, and chapters
HTML (.html)	Readable export with chapter sidebar and timestamp links
PDF (.pdf)	Printable document with chapters and timestamps

---

Publishing Desktop Releases

Release builds are fully automated via GitHub Actions. No local build needed.

git tag -a v1.0.4 -m "CapYap v1.0.4"
git push origin v1.0.4

GitHub Actions builds and publishes installers for all platforms to GitHub Releases.

Workflow: .github/workflows/desktop-release.yml

---

Project Structure

Path	Description
apps/backend/	FastAPI + LangGraph agent backend
apps/frontend/	React/Vite UI
apps/desktop/	Tauri desktop shell (macOS/Windows/Linux)
src/youtube_video_summarizer/	Python package and CLI entrypoints

Docs

• docs/README.md
• docs/AGENTIC_SYSTEM_ARCHITECTURE.md
• docs/DEVELOPMENT_GUIDE.md
• docs/SECURITY_AND_PRIVACY.md
• docs/OLLAMA_SETUP.md
• docs/RELEASE_CHECKLIST.md
• docs/CAPYAP_CLI.md
• apps/UI_LLM_BRIEF.md

CLI Tools (Optional)

CapYap includes script-friendly CLI commands: yt-extract-summarize and yt-transcript-chat. See docs/CAPYAP_CLI.md.

Troubleshooting

• capyap: command not found — Ensure env is active: conda activate capyap, then pip install -e .
• Frontend build not found — Run once from repo root: npm run build
• Port already in use — Desktop app may already be running backend on 127.0.0.1:8000. Use --port 8080 or connect to the existing instance.

Contributing

See CONTRIBUTING.md.

License

MIT