Who autoclip is for#
Content creators repurposing long-form interviews and podcasts
Paste a Bilibili or YouTube link, let autoclip score the segments, and export a set of short clips in one session instead of scrubbing through the recording manually. The export presets for YouTube Shorts and Bilibili remove the platform-specific export step.
Skip if:
If the recording has no usable dialogue track (music, gameplay without clear speech structure), the subtitle-based scoring will produce weak results.
Educators and course creators extracting lecture highlights
Course recordings and lecture replays with dense spoken content score well in autoclip's analysis. The topic segmentation groups related explanations together, making it easier to identify clips that stand alone as content without additional context.
Skip if:
If your course relies heavily on diagram walkthroughs or screen demonstrations with minimal narration, clip quality may be inconsistent.
Developers building automated video clipping pipelines
The CLI accepts local files and outputs JSON results, which fits standard batch ETL patterns. The MCP server exposes the full processing pipeline to MCP-compatible clients, allowing AI agents to trigger autoclip analysis programmatically. Docker Compose deployment makes it straightforward to run as a background service.
Skip if:
If you need frame-level visual analysis or facial recognition to drive clip selection, autoclip's subtitle-only approach is not the right fit.
Self-hosters replacing OpusClip or Klap for cost reasons
For teams or individuals currently paying monthly subscriptions to OpusClip, Vizard, or Klap, autoclip removes the recurring cost. Running on a modest server or a personal machine handles most typical workloads. The MIT license permits commercial use without additional licensing fees.
Skip if:
If you process a high volume of videos and need managed infrastructure, uptime guarantees, or transcription quality above what local Whisper provides, a managed service remains the more operationally sound choice.
The problem it solves#
Creating short clips from long videos is time-consuming by hand. Watching an hour-long recording to find the five minutes worth sharing, then cutting each clip, trimming handles, adding title cards, and exporting at the right resolution for each platform can easily take longer than the original recording. For creators who publish weekly, this editing burden compounds fast.
Paid clipping tools solve the time problem but introduce a different one. Your footage gets uploaded to someone else's server, analyzed by their model, and returned as clips you have no control over. Subscription costs add up, model quality is fixed to whatever the service decides, and if the platform changes pricing or shuts down, your workflow goes with it.
How it solves it#
Local-first video processing
Video files are processed entirely on your machine. Only subtitle text is sent to the language model you select; the raw video never leaves your device. This applies to all three deployment modes: desktop, Docker, and CLI.
Multi-source import with local Whisper transcription
Accepts Bilibili links, YouTube links, and local MP4/MOV/MKV files. Videos without subtitles can be transcribed locally using Whisper, installed in one step from the settings page. Existing SRT files can be provided at import time.
Bring-your-own model support
Supports Qwen (Tongyi Qianwen), OpenAI-compatible APIs, Google Gemini, Siliconflow, and local models via Ollama or LM Studio. API keys stay in your own settings page. You can switch providers per-project or point at a custom OpenAI-compatible base URL.
Automated clip scoring and collection assembly
The language model reads the full transcript, segments it by topic, and assigns a highlight score to each segment. Clips above the scoring threshold are exported automatically. You can adjust the threshold, reorder clips in the collection view, and edit titles before export.
Export presets for major platforms
Built-in export presets cover YouTube Shorts, TikTok, Xiaohongshu, and Bilibili. The exporter can burn subtitles and add title cards. Multi-account Bilibili upload is supported from the desktop app.
Three deployment modes: desktop, Docker, and CLI/MCP
The desktop app (macOS Apple Silicon, Windows) bundles Python and FFmpeg with no environment setup. The Docker Compose stack provides a web interface for server or NAS use. The CLI handles batch automation; the MCP server exposes the same pipeline to any MCP-compatible client.
Strengths and trade-offs#
Strengths
- MIT license with no video upload requirementautoclip is MIT licensed, meaning you can run it commercially, modify it, or fork it without restriction. Unlike OpusClip or Klap, which upload your video to their own servers for processing, autoclip keeps footage on your device. Only subtitle text leaves your machine.
- Zero-dependency desktop installerThe desktop package bundles a portable Python runtime and a static FFmpeg binary. There is no need to install Python, configure a virtual environment, or locate a working FFmpeg build. Download the DMG or exe, open it, and the tool is ready to use.
- Model provider freedom, including local modelsYou are not bound to a single AI provider. Qwen, OpenAI-compatible APIs, Gemini, Siliconflow, Ollama, and LM Studio all work. Swapping providers changes no other part of the workflow, and a custom base URL lets you connect to any OpenAI-compatible endpoint, including private deployments.
- Active development with strong early community tractionThe repository reached 8,243 stars and 1,572 forks since its July 2025 launch, appeared on GitHub Trending, and pushed a release as recently as September 2026. The 16 open issues indicate an active but manageable queue for a project of this age.
Trade-offs
- -Desktop app limited to Apple Silicon Macs and Windows x64The prebuilt desktop installer runs on macOS with Apple Silicon (M-series) and Windows 10/11 x64. Intel Macs and Linux are not supported by the desktop app; users on those platforms need the Docker or CLI path. The Docker path requires Docker Compose v2 and familiarity with environment file configuration.
- -Subtitle-driven analysis works poorly on non-speech contentHighlight scoring is based on transcript text, not visual or audio signal. Content that is predominantly musical, action-based, or otherwise non-verbal will produce poor or empty clip results. The README recommends interviews, podcasts, courses, and talk recordings as the primary use cases.
- -App packages lack code signingThe macOS DMG is not Apple-notarized and the Windows installer is not code-signed. On first launch, macOS Gatekeeper and Windows SmartScreen both display a security warning. Users must right-click to open on macOS or click 'More info' on Windows. Code signing is noted as planned but not yet shipped.
- -Solo-maintained with no guaranteed support timelineThe project is maintained by a single developer in their spare time. Response times on GitHub Issues and Discussions are not fixed, and there is no paid support tier or SLA. For production or team workflows that depend on timely bug fixes, this is a meaningful constraint.
autoclip vs alternatives#
autoclip vs OpusClip
OpusClip is a managed cloud service for AI-powered video clipping, while autoclip is a self-hosted open source tool. Both accept YouTube links and analyze content to extract highlights, but the underlying approach differs on data handling and cost structure.
| Feature | autoclip | OpusClip |
|---|---|---|
| License | MIT | Proprietary |
| Video upload | None (local processing) | Required (cloud) |
| Self-hosting | Yes | No |
| Pricing | Free (bring your own API key) | Paid subscription |
| Model choice | Open (Qwen, Gemini, OpenAI-compatible, local) | Fixed |
| Desktop app | Yes (macOS Apple Silicon, Windows) | Web only |
autoclip is the better choice when data privacy is the deciding factor, when you want to control which language model analyzes your content, or when you process enough videos that a monthly subscription is a meaningful cost. OpusClip is worth considering if you need a fully managed workflow with no infrastructure to maintain, or if your team needs built-in collaboration and sharing features.
autoclip vs Klap
Klap is a subscription-based clipping service targeting creators who repurpose long-form video into short clips. Like OpusClip, it handles processing in the cloud.
| Feature | autoclip | Klap |
|---|---|---|
| License | MIT | Proprietary |
| Video upload | None (local processing) | Required (cloud) |
| Self-hosting | Yes | No |
| Pricing | Free (bring your own API key) | Paid subscription |
| Export presets | YouTube Shorts, TikTok, Bilibili, Xiaohongshu | YouTube Shorts, TikTok, Instagram |
| CLI/API access | Yes (CLI + MCP server) | No public API |
autoclip's CLI and MCP server give it an automation edge that Klap does not offer. Developers who want to batch-process videos programmatically or integrate clipping into a larger pipeline will find autoclip more capable here. Klap's polished interface and managed infrastructure remain the better fit for non-technical creators who prioritize ease of use over server management.
Quick start#
Self-hosting autoclip with Docker Compose requires Docker and Docker Compose v2.
```bash
git clone https://github.com/zhouxiaoka/autoclip.git
cd autoclip
cp env.example .env
mkdir -p data logs uploads
docker compose up -d --build
```What it's built on#
- Languages
- JavaScriptPythonRustTypeScript
- Frameworks
- FastAPIReact
- Cache
- Redis
FAQ#
Does autoclip upload my videos to a server?
No. Video files are processed locally on your machine. The only data sent externally is subtitle text, which is forwarded to the language model provider you choose in your settings. If you use a local model via Ollama or LM Studio, nothing leaves your device at all.
Is autoclip free to use?
autoclip itself is free and MIT licensed. The only cost is the API usage from your chosen language model provider, which the README estimates at a few cents to fractions of a dollar per video depending on length and model. Using a local model via Ollama eliminates API costs entirely.
Which platforms does autoclip export to?
autoclip includes export presets for YouTube Shorts, TikTok, Xiaohongshu, and Bilibili. It can burn subtitles into clips and add title cards. Multi-account Bilibili upload is available from the desktop app.
What types of video content work best with autoclip?
autoclip analyzes transcript text to score segments, so it works best with speech-heavy content: interviews, podcasts, courses, and live stream recordings with substantive spoken discussion. Purely visual content, music videos, or gameplay footage without substantial dialogue will produce weaker highlight selections.
How do I run autoclip on Linux or an Intel Mac?
The desktop app does not support Linux or Intel Macs. Both platforms can run autoclip via Docker Compose, which provides the same web interface as the desktop app, or via the CLI using Python 3.10+ and FFmpeg installed in a virtual environment.
Similar open-source tools#
rakazo
AI teammates you own: your keys, your model, your machine.
Codex-X
Visual desktop companion for OpenAI Codex configuration
OpenSpec
Spec-driven development for AI coding assistants
tradingview-mcp
MCP bridge for AI-assisted TradingView chart analysis
cli
Official Lark/Feishu CLI with 200+ commands and AI Agent Skills
Multica
Assign GitHub issues to AI coding agents. They run on your machine.