Broedkrummen

brave-api-search

知识管理v4.2.1
作者:Broedkrummen更新于 2026-10-03共 13 个版本 · 最新版本发布于 2026-10-03@clawhub_broedkrummen/brave-api-search
信息检索#Documentation#Web Search

据 AlphaMoat 统计,截至 2026-10,brave-api-search 累计下载 12,814 次、获 2 个星标,按累计下载量在本次参与排名的 10,914 个 AI 技能中排名第 2,177 位,在「知识管理」分类(745 个)中排名第 143 位。

列表默认按累计下载量排序,也可按星标排序;详情页排名按累计下载量计算。

累计下载
1.3万
星标
2
全站下载量排名
#2,177
全站前 20% · 「知识管理」前 20%
平台累计下载星标评论
ClawHub ↗1.2万2-
SkillHub ↗1.3万2—

各平台的计数口径不同,统计时间不一致。

这个技能是做什么的

通过 Brave API 使用搜索与回答密钥,建议和拼写可另配专用密钥。搜索返回标题、链接、说明;回答模式可综合来源并提取实体。文档对非流式引文描述冲突,因此需要引文或实体时使用流式并核验返回来源。查询发送给 Brave,需保护密钥并关注额度。

资料发布于 2026-10-27

可以用来做什么

  • 结合网页链接研究主题

什么时候用

  • 需要当前外部信息或查询辅助时

示例提示词

以下是根据技能文档生成的示例,不代表实际运行结果。

搜索这个主题,核对引文后总结来源。

月度数据趋势

累计下载1.3万
8,8721.1万1.3万2026-05 · 累计下载 9,1642026-06 · 累计下载 9,5062026-07 · 累计下载 1.1万2026-08 · 累计下载 1.2万2026-09 · 累计下载 1.2万2026-10 · 累计下载 1.3万26-0526-0926-10
星标2
1232026-05 · 星标 22026-06 · 星标 22026-07 · 星标 22026-08 · 星标 22026-09 · 星标 22026-10 · 星标 226-0526-0926-10

2026-09 起统计方法调整:同名技能按作者分别统计,并覆盖全部技能。图中虚线为调整前的月份,与之后的数据不直接比较,也不计算跨越调整的增长。

月份累计下载环比星标
2026-101.3万—2
2026-091.2万—2
2026-081.2万+9.1%2
2026-071.1万+11.4%2
2026-069,506+3.7%2
2026-059,164—2

SKILL.md 原文

以下是作者发布的原始文件,按纯文本显示,本站未做修改或审核;其中的链接不可点击,命令请核实后再运行。抓取于 2026-10-23。

10 KB · ClawHub v4.2.1
---
name: brave-api-search
description: Real-time web search, autosuggest, and AI-powered answers using the official Brave Search API. Use for searching documentation, facts, current events, or any web content. Supports AI grounded answers with citations and query autosuggest. Requires BRAVE_SEARCH_API_KEY and BRAVE_ANSWERS_API_KEY.
license: MIT
metadata:
  author: Broedkrummen
  version: 4.2.1
---

# Brave API Search v4.2.0 — Significant Update
展开全部(共 227 行)

## Native OpenClaw plugin

The same four tools are now available as a native plugin: [Brave API Search plugin](https://clawhub.ai/plugins/brave-api-search-plugin). Install it with `openclaw plugins install clawhub:brave-api-search-plugin`. This skill remains available for existing users.


Real-time web search, autosuggest, spellcheck, and AI-powered answers using the official Brave Search API. Four tools:
- `brave_search` — web results with titles, URLs, descriptions, optional AI summary
- `brave_suggest` — query autosuggestions as users type with optional rich metadata
- `brave_spellcheck` — spell correction and "Did you mean?" suggestions
- `brave_answers` — AI-grounded answers with inline citations powered by live web search
  | **NEW — v4.2.0:** Full entity extraction via `enable-entities` flag

## Setup

Set your Brave API keys in a local `.env` file (recommended):

```bash
# .env (do not commit)
BRAVE_SEARCH_API_KEY=your_key_here
BRAVE_ANSWERS_API_KEY=your_key_here
AUTOSUGGEST_API_KEY=your_key_here   # optional, for autosuggest only
BRAVE_SPELLCHECK_API_KEY=your_key_here   # optional, for spellcheck only
```

Get your keys at: https://api-dashboard.search.brave.com

**Key routing:**
- `brave_search` → `BRAVE_SEARCH_API_KEY`
- `brave_answers` → `BRAVE_ANSWERS_API_KEY`
- `brave_suggest` → `AUTOSUGGEST_API_KEY` → `BRAVE_AUTOSUGGEST_API_KEY` → `BRAVE_SEARCH_API_KEY` (tries in order)
- `brave_spellcheck` → `BRAVE_SPELLCHECK_API_KEY` → `BRAVE_SEARCH_API_KEY` (tries in order)

> Note: `brave_answers` uses streaming by default. Streaming is required for citations, entities, and research mode.

## When to Use This Skill

**Use `brave_search` when:**
- Searching for current information, news, or recent events
- Looking up documentation or technical references
- Need ranked results with URLs to follow up on
- Want an AI summary of search results

**Use `brave_suggest` when:**
- Power autocomplete in search interfaces
- Help users formulate better queries faster
- Need query completions as users type
- Want rich metadata (titles, descriptions, images) for suggestions

**Use `brave_answers` when:**
- Need a synthesized answer with cited sources
- Researching topics that benefit from multiple sources
- Want AI-grounded responses with inline citations
- Deep research mode needed (multi-search)

**Don't use this skill for:**
- Questions already answered from context or memory
- Tasks that don't require external information

## Tools

### brave_search

Web search returning ranked results with titles, URLs, and descriptions.

```bash
brave_search --query "latest Node.js release" --count 5
brave_search --query "TypeScript generics" --extra-snippets true
brave_search --query "current weather Copenhagen" --freshness pd
brave_search --query "React Server Components" --summary true
```

**Parameters:**
- `query` (required) — Search query, supports operators: `site:`, `"exact phrase"`, `-exclude`
- `count` — Results to return (1-20, default: 10)
- `country` — 2-letter country code (default: `us`)
- `freshness` — Date filter: `pd` (24h), `pw` (7 days), `pm` (31 days), `py` (1 year)
- `extra-snippets` — Include up to 5 extra text excerpts per result (default: false)
- `summary` — Fetch Brave AI summarizer result (default: false)
- `offset` — Pagination offset for next page results

**Returns:** Formatted list of results with title, URL, description, optional AI summary, and a hint for next page if `more_results_available`.

### brave_suggest

Query autosuggest API providing intelligent query autocompletion as users type.

```bash
brave_suggest --query "hello"
brave_suggest --query "pyt" --count 5 --country US
brave_suggest --query "einstein" --rich true
```

**Parameters:**
- `query` (required) — Partial query to get suggestions for
- `count` — Number of suggestions (1-10, default: 5)
- `country` — 2-letter country code (default: `US`)
- `rich` — Include enhanced metadata: titles, descriptions, images, entity detection (default: false, requires paid Autosuggest plan)

**Returns:** List of query suggestions, optionally with rich metadata. Results are cached for 60 seconds.

**Best Practices:**
- Implement debouncing (150-300ms) to avoid excessive API calls as users type
- Load suggestions asynchronously without blocking the UI

### brave_spellcheck

Spell checking and query correction for search queries.

```bash
brave_spellcheck --query "articifial inteligence"
brave_spellcheck --query "hello" --country US
```

**Parameters:**
- `query` (required) — Query to check for spelling errors
- `country` — 2-letter country code (default: `US`)

**Returns:** "Did you mean:" suggestion, or confirms no correction needed. Uses dedicated `BRAVE_SPELLCHECK_API_KEY` if available, falls back to `BRAVE_SEARCH_API_KEY`.


### brave_answers

AI-powered answers grounded in live web search with inline citations.

```bash
brave_answers --query "How does React Server Components work?"
brave_answers --query "Compare Postgres vs MySQL for OLAP" --enable-research true
brave_answers --query "Latest Python release notes" --enable-citations true
brave_answers --query "Who is Albert Einstein" --enable-entities true
```

**Parameters:**
- `query` (required) — Question or topic to research
- `country` — Target country for search context (default: `us`)
- `enable-citations` — Include inline source citations (default: true)
- `enable-research` — Multi-search deep research mode (default: false)
- `enable-entities` — Include entity information in responses (default: false, streaming required)
- `stream` — Enable streaming output (default: true — required for citations/entities/research)

**Returns:** AI answer with cited sources extracted from the response, plus token usage and cost breakdown.

> ⚠️ **Citations, entities, and research mode require streaming (`--stream true`, which is the default).** If you disable streaming with `--stream false`, citations and entity data will not be available.

**Streaming mode output:**
- Text streams progressively to stdout
- Entities streamed inline as `<enum_item>` tags are parsed
- **Entities mentioned** section printed at the end (v4.2.0)
- Sources printed at the end (sorted by citation number)
- Token usage and cost breakdown included

**Non-streaming fallback:** Use `--stream false` for simpler output at the cost of losing citations and entity data.

## Pricing & Limits

Brave pricing is credit-based and can change. Do **not** assume a fixed free request count.

Current public guidance (verify in Brave dashboard/docs before production use):
- Monthly trial credits may be offered (e.g. `$5 in monthly credits`)
- Search and Answers consume credits differently
- Rich suggestions require a paid Autosuggest plan
- Answers may also include token-based costs
- QPS limits depend on your plan tier

Always check your live limits and usage in:
- https://api-dashboard.search.brave.com
- https://brave.com/search/api/

## Technical Details

### Architecture

The skill uses a shared `utils.js` module containing:
- `parseArgs()` — CLI argument parser
- `fetchWithRetry()` — fetch with exponential backoff retry on 429 and 5xx
- `createCache()` — in-memory TTL cache (used by `brave_suggest`)

### brave_search.js

The core search script with the following optimizations:
- **Parallel summary fetch:** When `--summary` is enabled, the summary key is extracted from the search response and fetched in parallel with the search results
- **Pagination hint:** Response includes a hint for the next page offset when `more_results_available` is true
- **Rate limit handling:** Retries on 429 (honors `Retry-After` header) and 5xx (exponential backoff)

### brave_answers.js v4.2.0

The AI answers script supports two modes:
- **Streaming (default):** Progressive output with real-time entity extraction, deduplicated citations, token usage, and cost breakdown
- **Non-streaming:** Full citation and entity extraction from batch response via `parseCitations()` and `parseEntities()`

Key changes in v4.2.0:
- **Entity extraction:** Streaming mode parses `<enum_item>` tags incrementally — entity names are streamed inline as they are parsed, and a consolidated **Entities mentioned** section is printed at the end with URL and citation count per entity
- **Citation deduplication:** Duplicate source URLs are deduplicated so each source is cited only once (sorted by citation number)
- **Non-streaming mode (v4.2.0):** Full `parseCitations()` and `parseEntities()` applied to the batch response content, so citations and entities are extracted even without streaming

### brave_suggest.js

The autosuggest script with the following optimizations:
- **In-memory cache:** 60-second TTL cache keyed by `query|count|country|rich` — deduplicates rapid successive calls
- **Key fallback chain:** `AUTOSUGGEST_API_KEY` → `BRAVE_AUTOSUGGEST_API_KEY` → `BRAVE_SEARCH_API_KEY`

### brave_answers streaming

When streaming is enabled (`stream: true`):
- SSE chunks are parsed in real-time
- Citations are collected and printed at end (sorted by citation number)
- Entity items (`<enum_item>`) are parsed incrementally — streamed inline and collected for the **Entities mentioned** section
- Usage block (`<usage>`) parsed and displayed with token counts and total cost

## Security & Packaging Notes

- This skill only calls Brave official endpoints under `https://api.search.brave.com/res/v1`.
- It requires `BRAVE_SEARCH_API_KEY` and `BRAVE_ANSWERS_API_KEY` (keep them in `.env`, not inline in commands/chats).
- Optionally accepts `AUTOSUGGEST_API_KEY` or `BRAVE_AUTOSUGGEST_API_KEY` for autosuggest.
- It does not request persistent/system privileges and does not modify system config.
- All scripts are Node.js 18+ native (no external dependencies).

「知识管理」分类下的热门技能

搜索引擎知识管理

通过公开搜索URL在16个搜索引擎(7个国内直连+9个国际)中做跨引擎检索、结果去重与多源交叉验证,无需API Key,内置零依赖Python脚本。Use when 用户要求“搜一下/查资料/多找几个来源/核实消息/找公众号文章/排查报错/找论文或开源项目”时。输入自然语言查询词,输出带来源链接与可信度标注的结构化结果。不适用于私有数据检索、批量抓取、需要登录后可见的内容。

↓ 1075.9万★ 147u_8ee34d77
ima-skills知识管理

通过 IMA OpenAPI 管理笔记与知识库内容。

↓ 108.4万★ 566u_1742680f
Summarize知识管理

把长文本、文档与网页内容压缩成要点和关键词的摘要工具。

↓ 87.6万★ 1,048paudyyin
全网搜索知识管理

通过“查一下、搜一下、搜索、找一下”等自然语言检索公开网页,整理来源并交付横版 PDF、Markdown 与标题链接表。支持用户主动开启、查看和取消技能收藏,保留原报告取回。

↓ 50.9万★ 2u_32d99f8a
web-tools-guide知识管理

规定如何为网页请求选择合适工具的必读指引。

↓ 40.0万★ 220user_ec205dbb
Aihot Skill Lite知识管理

AI HOT(aihot.virxact.com)中文 AI 资讯查询 Skill,帮助用户获取今日 AI 圈动态、AI 日报、AI 热点、最新 AI 资讯以及 OpenAI、Anthropic、Google 等公司的最新发布信息。

↓ 32.1万★ 92Khazix
全球12亿文献知识库(8千万中文期刊可下载)知识管理

全球12亿文献知识库(8千万中文期刊可下载)——通过 SmartLib 开放平台 API 提供中外文学术文献检索与下载能力,覆盖 8000 万篇授权中文期刊全文 + 12.28 亿条全球文献元数据(期刊 7.19 亿 / 专利 2.15 亿 / 会议 7155 万 / 学位论文 2473 万 / 标准 268 万)。 能力:中英文关键词检索、文献详情、中文期刊 PDF 全文下载、外文 OA 文献十级渠道免费下载(不消耗配额)、智能关键词扩展、核心期刊优先排序、相关性重排、引文追溯、分类号检索。 配额:首次使用自动注册,免费 100 次检索 + 10 次下载 / 月;耗尽自动弹出套餐(体验卡 / 个人版月 / 专业版月 / 单篇下载 / 下载包),可说「升级 / 充值」唤起;企业 / 机构定制联系 [email protected]。 触发:用户表达"查论文""找文献""检索学术""搜期刊""查专利""找标准""下论文""写文献综述""找参考文献""查 SCI/EI"等意图时启用;也适用"帮我找关于 XX 的论文""写文献综述""找几篇引用支撑论点"。英文:"find papers" "search literature" "write literature review" "find supporting citations"。 调用前必须先用 /consume 获取 consume_token,再凭 token 调 /search(每次计费接口调用都需一次 /consume)。

↓ 30.3万★ 189user_164f4c1f
Markdown Converter知识管理

用 uvx 直接调用 markitdown,把各类文件转成 Markdown。

↓ 20.4万★ 256Peter Steinberger

数据月份:2026-10 · 下载/星标数聚合自公开技能注册库(ClawHub、SkillHub),按月更新。口径说明见数据方法。