OpenClaw Can’t Handle Bilibili or Xiaohongshu Natively? bibigpt-skill Fixes It in One Command

Source (Chinese original): https://bibigpt.co/blog/posts/openclaw-bibigpt-skill-ai-agent-video-2026

This is an English translation and adaptation for ClawList readers.

In 2026, OpenClaw became one of the hottest AI agent runtimes, crossing 180K+ GitHub stars. It can automate shell tasks, messages, calendars, and workflows.

But for Chinese-speaking users, one practical gap remains obvious:

OpenClaw’s native summarize does not support Bilibili, Xiaohongshu, or Douyin.

That gap is exactly what bibigpt-skill solves.

---

5-Minute Quick Start

1. Install BibiGPT Desktop (macOS/Windows)
2. Install skill:

npx skills add JimmyLv/bibigpt-skill

3. Verify login:

bibi auth check

4. In Claude Code / OpenClaw, ask:

Summarize this video: <url>

(Bilibili + YouTube both supported)

---

Why OpenClaw Exploded in 2026

OpenClaw marks a shift from “answering questions” to “executing tasks.”

Core strengths:

• Shell execution and file operations
• External system integration (Telegram/Slack/email/etc.)
• Heartbeat automation for scheduled autonomous tasks
• Large skill ecosystem

A key 2026 milestone is spawn mode with coding agents: OpenClaw handles planning and orchestration, and a coding agent handles execution.

bibigpt-skill plugs into this ecosystem as the video understanding layer.

---

The Video Blind Spot of AI Agents

Most agent workflows are strong with text/code/APIs, but weak with video—especially Chinese platforms.

Where important Chinese content lives:

• Bilibili: tutorials, deep dives, long-form explainers
• Xiaohongshu: product reviews, lifestyle workflows, practical notes
• Douyin: short-form trends and tactical updates
• Podcasts: long-form audio insights

Without specialized tooling, agents can only pass around links instead of extracting knowledge.

bibigpt-skill exposes BibiGPT’s Chinese-platform-optimized pipeline as a standard CLI, so any shell-capable agent can use it.

---

What bibigpt-skill Adds

Prerequisite

# macOS
brew install --cask jimmylv/bibigpt/bibigpt

# Windows
# Download desktop app from: https://bibigpt.co/download/desktop

Install

npx skills add JimmyLv/bibigpt-skill

Verify

bibi auth check
bibi --help

Command Cheat Sheet

bibi summarize "<url>"                # standard summary
bibi summarize "<url>" --chapter      # chaptered summary
bibi summarize "<url>" --subtitle     # transcript/subtitles only
bibi summarize "<url>" --json         # structured JSON output
bibi summarize "<url>" --async        # async processing for long videos

---

Native summarize vs bibigpt-skill

| Capability | OpenClaw native summarize | bibigpt-skill | |---|---:|---:| | YouTube | ✅ | ✅ | | Bilibili | ❌ | ✅ | | Xiaohongshu | ❌ | ✅ | | Douyin | ❌ | ✅ | | Podcasts | ❌ | ✅ | | Local audio/video files | ❌ | ✅ | | Chaptered summary | ❌ | ✅ (--chapter) | | Structured JSON | ❌ | ✅ (--json) | | Transcript extraction | ❌ | ✅ (--subtitle) | | Async long-video processing | ❌ | ✅ (--async) |

Native summarize is a general web-summary tool. bibigpt-skill is a platform-specific pipeline optimized for real-world video extraction + structuring.

---

3 Practical Workflows

Path 1: Claude Code Daily Workflow

Use directly in your coding session:

Summarize this video and extract implementation details: <url>

Good for:

• learning from technical talks
• competitor feature analysis
• digesting product demos and recordings

Path 2: OpenClaw → Coding Agent → bibigpt-skill

OpenClaw (planning) -> coding agent (execution) -> bibigpt-skill (video understanding) -> structured report

Good for:

• competitor intelligence
• conference/video tracking
• content research pipelines

Path 3: OpenClaw Heartbeat Automation

Daily autonomous video intelligence:

1. Pull target channel feeds
2. Detect videos from past 24h
3. Run bibi summarize --chapter --json
4. Aggregate into a markdown brief
5. Send to Slack/email

This turns passive consumption into proactive, repeatable research.

---

Beyond Summary: Why BibiGPT Pipeline Matters

bibi summarize is only one entry point. The broader stack supports capabilities like timestamp-grounded QA and source tracing, helping reduce hallucination risk in downstream AI workflows.

It can also generate highlight notes with time anchors, which can be reused in daily briefs, knowledge bases, and agent memory systems.

---

Why This Skill Is Important

OpenClaw proved that AI-agent value scales with skill ecosystem depth.

Video remains a major “knowledge island.” A bridge like bibigpt-skill allows agents to understand, extract, and operationalize video knowledge—not just share URLs.

For Chinese internet workflows in particular, this closes a critical capability gap.

---

Start Now

# 1) Install desktop app (macOS)
brew install --cask jimmylv/bibigpt/bibigpt

# 2) Install skill
npx skills add JimmyLv/bibigpt-skill

# 3) Verify
bibi auth check

# 4) Test in agent session
# "Summarize this video: https://..."

• Skill repo: https://github.com/JimmyLv/bibigpt-skill
• Website: https://bibigpt.co
• App: https://bibigpt.co/app
• Desktop: https://bibigpt.co/download/desktop
• Features: https://bibigpt.co/features

---

Translator note: This translation preserves the technical structure and key claims of the original article while adapting phrasing for ClawList’s English audience.