Search Skills
Three search skills: four-source search scored by intent, deep thread tracing and clean Markdown extraction from pages.
- 438 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Scripts make network calls and send queries and links to external search services and the MinerU API.
- Exa, Tavily and Grok keys live in a local file or environment variables.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/search-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Scripts make network calls and send queries and links to external search services and the MinerU API.
- Exa, Tavily and Grok keys live in a local file or environment variables.
This is third-party code. Review the repository files before installing.
What it does
A set of three skills for agents. `search-layer` searches Brave, Exa, Tavily and Grok in parallel, classifies the query intent (fact, status, comparison, tutorial, exploration, news, resource), removes duplicates and scores results by freshness, authority and keyword match. Research queries get an extra stage with Exa deep search. The skill traces discussion threads deeper: it reads GitHub issues and PRs, Hacker News, Reddit and V2EX posts with their comments and references. `content-extract` turns a URL into clean Markdown and, when a page such as WeChat fails, falls back to `mineru-extract`, a wrapper over the official MinerU API that reads PDFs, Office documents and HTML.
Who it is for. Researchers and developers who ask an agent to gather sources on a topic and trace what an issue discussed, rather than take the first search results.
Good fit when
- A question about the current state of a project needs fresh sources from several systems.
- A bug cause must be found by following links from an issue to PRs and commits.
- Two tools need comparing through articles and Hacker News and Reddit threads.
- A WeChat article does not open with an ordinary fetch and Markdown text is needed.
Not a fit when
- Exa and Tavily keys are unavailable: without them four sources shrink to one.
- The data is confidential and queries cannot go to external search services.
- A single quick web query is enough without scoring and deduplication.
Example request
Find what is known about the OpenClaw config validation bug, follow the links from the issue and show the sources.Limitations
The README and skill instructions are mostly in Chinese and installation is described for OpenClaw: in other agents the skill folders are wired in by hand. `search-layer` needs your own Exa and Tavily keys, Grok is optional, and Brave comes through the agent's built-in `web_search`. `mineru-extract` needs a MinerU token. Queries and page contents go to the chosen providers. The scripts are Python 3.10 or newer and run on your machine.
How to disable. Remove the search-layer, content-extract and mineru-extract links from the agent's skills directory and the keys from ~/.openclaw/credentials/search.json.
FAQ
Which search sources does search-layer use?
Brave through `web_search`, plus Exa, Tavily and Grok. Without Grok configured, search runs on Exa and Tavily.
How does the skill understand query intent?
It assigns the query to one type: fact, status, comparison, tutorial, exploration, news or resource. The type sets the mode, freshness window and scoring weights.
What does content-extract do on protected pages?
It tries ordinary extraction first and hands the page to `mineru-extract` on failure. The result comes back as Markdown with a link to the source.
Which platforms can it trace in depth?
GitHub issues and PRs, Hacker News, Reddit, V2EX and ordinary web pages. The `fetch_thread.py` script takes one address whole.
Which keys does search-layer need?
Exa and Tavily keys are required and a Grok key is optional. They go into `~/.openclaw/credentials/search.json` or the `EXA_API_KEY` and `TAVILY_API_KEY` environment variables.
Discussion
No questions or reviews yet.
Related
SkillsA web research skill: SearXNG searches, Camofox reads pages and CloakBrowser opens sites that check their visitors
Six OpenClaw skills: four-source search, page extraction to Markdown, GitHub repository analysis and dependency checks.
Official Brave skills that teach an agent to call the Search API for web, news, images, videos, places and page extracts for LLMs
Two small skills: web search through your own browser with no paid API, and a preflight risk check before a meaningful change.
Three skills and four utilities that let an agent search the web, read pages, pull video subtitles and process media
A skill for web search, X search and page fetching for models without web access: no signup or key, the answer is JSON with source links
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Search Skills (this card) | no | 438 | |
| Tavily Agent Skills | yes | 490 | skills from a single provider, Tavily; here four sources share scoring and discussion thread tracing |
| Exa Research MCP Skills | no | 118 | relies on Exa; search-layer merges Exa, Tavily, Grok and Brave and removes duplicates |
README in short
There are two paths: main retrieval with scoring and deduplication, and a thread-pulling path along references. Both share one principle: find the right information first, then decide whether to dig deeper. The repository is also part of the aggregate openclaw-skills collection.
README badge
Original SKILL.md
---
name: search-layer
description: >
DEFAULT search tool for ALL search/lookup needs. Multi-source search and deduplication
layer with intent-aware scoring. Integrates Brave Search (web_search), Exa, Tavily,
and Grok to provide high-coverage, high-quality results. Automatically classifies
query intent and adjusts search strategy, scoring weights, and result synthesis.
Use for ANY query that requires web search — factual lookups, research, news,
comparisons, resource finding, "what is X", status checks, etc. Do NOT use raw
web_search directly; always route through this skill.
---
# Search Layer v2.2 — 意图感知多源检索协议
四源同级：Brave (`web_search`) + Exa + Tavily + Grok。按意图自动选策略、调权重、做合成。
## 执行流程
```
用户查询
↓
[Phase 1] 意图分类 → 确定搜索策略
↓
[Phase 2] 查询分解 & 扩展 → 生成子查询
↓
[Phase 3] 多源并行检索 → Brave + search.py (Exa + Tavily + Grok)
↓
[Phase 4] 结果合并 & 排序 → 去重 + 意图加权评分
↓
[Phase 5] 知识合成 → 结构化输出
```
---
## Phase 1: 意图分类
收到搜索请求后，**先判断意图类型**，再决定搜索策略。不要问用户用哪种模式。
| 意图 | 识别信号 | Mode | Freshness | 权重偏向 |
|------|---------|------|-----------|---------|