---

Plug Big-Tech Native Search into Your OpenClaw

AI Search Hub is an open-source skill that turns fragmented AI search surfaces, page-extraction abilities, and the native data worlds behind those platforms into one reusable orchestration hub.

It is not just for one-query, all-search workflows. Give it a link and it can also borrow the page understanding, search, and extraction abilities already refined by major AI platforms. More importantly, it lets you reach data that those platforms are better positioned to surface, such as WeChat Official Accounts, Douyin, Weibo, and other native content ecosystems, then let them collect, clean, and structure the result for your agents and workflows.

It is not trying to make you maintain:

• fragile crawler stacks and brittle parsing rules
• separate browser automation and login recovery flows for every provider
• endless login, captcha, rate-limit, and anti-bot recovery
• direct work against fragmented entry points like WeChat, Douyin, Weibo, and the open web
• link-by-link body extraction, ad removal, noise cleanup, key-point extraction, and structuring by hand
• manual stitching of scattered results at the end

Its job is simple:

Give it a question and let those platforms search for you. Give it a link and let them fetch, read, clean, and structure it for you. The hardest data entry points are exactly where you let major platforms do the work for your agent.

Real Grok run result

Real Doubao run result

Real MiniMax run result

---

Install

AI Search Hub is not a traditional SDK. It is a skill.

Give this repository to any skill-capable agent or tool such as OpenClaw, Claude Code, Codex, or Cursor, and let it install or mount the skill directly.

---

Traditional Workflow vs AI Search Hub

Traditional workflow	AI Search Hub
🔧 Build crawlers yourself You own scraping, parsing, and long-term maintenance	⚡ Reuse platform-native search Plug directly into mature search surfaces
🧩 One automation flow per provider Every new source adds one more integration to maintain	🚀 One query, multi-provider dispatch Use one entrypoint for multiple search providers
🛡️ Constant anti-bot handling Login, rate limits, and UI changes drain engineering time	🎯 Lean on existing platform entry points Reduce repeated recovery work and focus on orchestration
🔍 Endless keyword tuning Usable coverage often needs constant search tweaking	🧠 Reuse provider-optimized search logic Reuse ranking, retrieval, and interaction patterns
🧱 Manual result stitching You still need to unify scattered outputs afterward	📦 Unified output for agents and workflows Pull results back into one reusable output channel

---

Supported Platforms

These platforms are not side notes. They are the search surfaces AI Search Hub orchestrates first.

Currently integrated: Gemini / Grok / Doubao / Yuanbao / LongCat / Qwen / MiniMax / Kimi

Platform	Search Strength	Typical Coverage	Status
Gemini Google-first discovery	Google search web discovery knowledge lookup	Google public web knowledge sites
Grok Real-time social search	X / Twitter live signals trend discovery	real-time posts social chatter trending topics
Doubao Chinese trend sensing	Chinese understanding hot topics content summarization	Douyin Chinese content trending discussions
Yuanbao Chinese source supplement	Chinese supplement official account lookup source cross-checking	WeChat Official Accounts Chinese web public content
LongCat Chinese knowledge blend	Chinese knowledge industry context structured summaries	Chinese knowledge bases industry reports public information
Qwen / Tongyi Chinese web expansion	Chinese search web Q&A public content	Chinese web search surfaces open information
MiniMax Agent General Chinese assistant	Chinese chat general web tasks consumer queries	general Chinese prompts lifestyle questions broad assistance
Kimi Long-context exploration	long context document reading information structuring	long-form content research material public web
Perplexity Web-native answer engine	web search source citation fast summaries	public web news sites knowledge content
Claude Reasoning and synthesis	complex reasoning information synthesis long-form summarization	research material multi-source synthesis knowledge Q&A
Wenxin Baidu ecosystem expansion	Chinese search Baidu ecosystem public web	Chinese pages search results public sites
More Vertical Sources Extensible surface	continuous expansion new surfaces new providers	vertical communities industry sites niche content sources

---

How it works

1. Receive one query

The user or agent asks one question only once. No repeated rewriting per provider.

2. Dispatch across providers

The same query is sent to multiple providers so each one can search the ecosystem it knows best.

3. Reuse platform-native strengths

Gemini is strong at Google and web search, but is still an experimental integration here. Grok is strong at X / Twitter and real-time discovery. Doubao leans into Douyin and Chinese trends, Yuanbao leans into WeChat-adjacent supplementation, Qwen leans into the public web, and LongCat leans into Chinese knowledge and industry context.

4. Collect and normalize results

Outputs from multiple providers are pulled back into a single channel, ready for normalization, fusion, and workflow consumption.

5. Return to your workflow

The final output is meant for agents, research pipelines, monitoring systems, and automation workflows, not just browser tabs.

---

Search Worlds It Can Reach

By reusing platform-native search capabilities, AI Search Hub can indirectly reach these core search worlds:

🌐 Global Web Google, public websites, knowledge pages	⚡ Real-Time Social X / Twitter, Reddit, live discussions	🔥 Chinese Trends Weibo, Douyin, Chinese hot topics
📰 WeChat Ecosystem WeChat Official Accounts, source supplements	🧭 Overseas Signals overseas social media, public trend signals	🏮 Chinese Internet Chinese web pages, vertical sites, content ecosystems

Original list backup

• Google
• Weibo
• Douyin
• X / Twitter
• Reddit
• WeChat Official Accounts
• public web pages
• overseas social media
• Chinese internet ecosystems

The point is not to crawl every platform yourself. The point is:

Let platforms search the worlds they already know best.

---

Advanced Usage

Example Configuration

# agents/openai.yaml
interface:
  display_name: "AI Search Hub"
  short_description: "Run AI Search Hub browser scripts across supported AI platforms"
  default_prompt: "Use $ai-search-hub to run one of the repository chat sites with automatic Chrome debug startup and login waiting."

policy:
  allow_implicit_invocation: true

This is not a conceptual interface. It is the actual agent configuration snippet currently used in the repository. The runtime entrypoint is still scripts/run_web_chat.py.

---

Example Request

python3 scripts/run_web_chat.py \
  --site doubao \
  --prompt "Help me plan a Xinjiang travel route" \
  --output out/doubao_xinjiang_route.txt

python3 scripts/run_web_chat.py \
  --site grok \
  --prompt "Summarize Elon Musk's posts on X from the last 14 days, list them by date in descending order, and include links" \
  --output out/grok_musk_recent.txt

Example Output

Doubao output excerpt:

1. Northern Xinjiang classic loop, 7-10 days
Urumqi -> Heavenly Lake -> Keketuohai -> Burqin -> Five-Colored Beach -> Kanas -> Hemu -> Urho Ghost City -> Sayram Lake

2. Ili grassland and flower route, 6-8 days
Urumqi -> Sayram Lake -> Guozigou -> Huocheng lavender -> Tekes -> Kalajun -> Xiata -> Nalati -> Duku Highway

Grok output excerpt:

2026-03-15: Starlink now available in Kuwait
2026-03-15: Couldn't script it better
2026-03-15: SpaceX 24th anniversary related post

Overall, the recent pattern on X is mostly three buckets: xAI/Grok and X platform promotion, Starlink/SpaceX, and frequent political commentary.

MiniMax output excerpt:

Based on Bilibili's trending board, here is a recent hot-video summary:

Games:
2. Prism 2033 gameplay reveal and first PV
6. I regretted bringing my friend to the dam

Lifestyle / outdoors:
1. Brother Niu, am I going to die too...
3. Rescuing a fox on a snowy mountain, but with one million farmers

Trend summary:
Hottest theme: the snowy-mountain fox rescue series
Popular categories: game animation, social observation, science explainers, emotional music

---

FAQ

Is this a crawler framework? No. It is a search capability aggregation skill.

Do I still need browser automation for every provider? Reducing that burden is exactly what AI Search Hub is trying to do.

Do I still need to tune keywords constantly? Far less than traditional workflows, because AI Search Hub tries to reuse search systems already optimized by the platforms.

Why is this useful for agents? Because agents need search as a reusable capability, not a pile of brittle scripts.

Can I add more providers? Yes. The design is intentionally lightweight and extensible.

---

Contributing

Issues and PRs are welcome.

If you also believe the future of search should not be one more crawler stack, but one smarter way to use all existing AI search surfaces together, feel free to contribute.

---

Star History

---

If this project is useful to you, consider giving it a Star.

If you have ideas, feedback, or want to see more platforms integrated, open an Issue or send a PR.