OpenClaw Search 🔍
Intelligent search for autonomous agents. Powered by AIsa.
One API key. Multi-source retrieval. Confidence-scored answers.
Inspired by AIsa Verity - A next-generation search agent with trust-scored answers.
🔥 What Can You Do?
Research Assistant
CODEBLOCK0
Market Research
CODEBLOCK1
Competitive Analysis
CODEBLOCK2
News Aggregation
CODEBLOCK3
Deep Dive Research
CODEBLOCK4
Quick Start
CODEBLOCK5
🏗️ Architecture: Multi-Stage Orchestration
OpenClaw Search employs a Two-Phase Retrieval Strategy for comprehensive results:
Phase 1: Discovery (Parallel Retrieval)
Query 4 distinct search streams simultaneously:
- - Scholar: Deep academic retrieval
- Web: Structured web search
- Smart: Intelligent mixed-mode search
- Tavily: External validation signal
Phase 2: Reasoning (Meta-Analysis)
Use AIsa Explain to perform meta-analysis on search results, generating:
- - Confidence scores (0-100)
- Source agreement analysis
- Synthesized answers
CODEBLOCK6
Core Capabilities
Web Search
CODEBLOCK7
Academic/Scholar Search
CODEBLOCK8
Smart Search (Web + Academic Combined)
CODEBLOCK9
Tavily Integration (Advanced)
CODEBLOCK10
Explain Search Results (Meta-Analysis)
CODEBLOCK11
📊 Confidence Scoring Engine
Unlike standard RAG systems, OpenClaw Search evaluates credibility and consensus:
Scoring Rubric
| Factor | Weight | Description |
|---|
| Source Quality | 40% | Academic > Smart/Web > External |
| Agreement Analysis |
35% | Cross-source consensus checking |
|
Recency | 15% | Newer sources weighted higher |
|
Relevance | 10% | Query-result semantic match |
Score Interpretation
| Score | Confidence Level | Meaning |
|---|
| 90-100 | Very High | Strong consensus across academic and web sources |
| 70-89 |
High | Good agreement, reliable sources |
| 50-69 | Medium | Mixed signals, verify independently |
| 30-49 | Low | Conflicting sources, use caution |
| 0-29 | Very Low | Insufficient or contradictory data |
Python Client
CODEBLOCK12
API Endpoints Reference
| Endpoint | Method | Description |
|---|
| INLINECODE0 | POST | Web search with structured results |
| INLINECODE1 |
POST | Academic paper search |
|
/scholar/search/smart | POST | Intelligent hybrid search |
|
/scholar/explain | POST | Generate result explanations |
|
/search/full | POST | Full text search with content |
|
/search/smart | POST | Smart web search |
|
/tavily/search | POST | Tavily search integration |
|
/tavily/extract | POST | Extract content from URLs |
|
/tavily/crawl | POST | Crawl web pages |
|
/tavily/map | POST | Generate site maps |
Search Parameters
| Parameter | Type | Description |
|---|
| query | string | Search query (required) |
| maxnumresults |
integer | Max results (1-100, default 10) |
| as_ylo | integer | Year lower bound (scholar only) |
| as_yhi | integer | Year upper bound (scholar only) |
🚀 Building a Verity-Style Agent
Want to build your own confidence-scored search agent? Here's the pattern:
1. Parallel Discovery
CODEBLOCK13
2. Confidence Scoring
CODEBLOCK14
3. Synthesis
CODEBLOCK15
For a complete implementation, see AIsa Verity.
Pricing
| API | Cost |
|---|
| Web search | ~$0.001 |
| Scholar search |
~$0.002 |
| Smart search | ~$0.002 |
| Tavily search | ~$0.002 |
| Explain | ~$0.003 |
Every response includes usage.cost and usage.credits_remaining.
Get Started
- 1. Sign up at aisa.one
- Get your API key
- Add credits (pay-as-you-go)
- Set environment variable: INLINECODE12
Full API Reference
See API Reference for complete endpoint documentation.
Resources
OpenClaw Search 🔍
面向自主智能体的智能搜索。由 AIsa 驱动。
一个 API 密钥。多源检索。带置信度评分的答案。
灵感来源于 AIsa Verity —— 一个提供可信评分答案的下一代搜索智能体。
🔥 你能做什么?
研究助手
搜索 2024-2025 年关于 Transformer 架构的最新论文
市场调研
查找所有关于 2025 年第四季度 AI 初创公司融资的网络文章
竞品分析
搜索 RAG 框架的评测与对比
新闻聚合
获取关于量子计算突破的最新新闻
深度研究
结合网络和学术来源,对autonomous agents进行智能搜索
快速开始
bash
export AISAAPIKEY=your-key
🏗️ 架构:多阶段编排
OpenClaw Search 采用两阶段检索策略以获得全面结果:
第一阶段:发现(并行检索)
同时查询 4 个不同的搜索流:
- - Scholar:深度学术检索
- Web:结构化网络搜索
- Smart:智能混合模式搜索
- Tavily:外部验证信号
第二阶段:推理(元分析)
使用 AIsa Explain 对搜索结果进行元分析,生成:
- - 置信度评分(0-100)
- 来源一致性分析
- 综合答案
┌─────────────────────────────────────────────────────────────┐
│ 用户查询 │
└─────────────────────────────────────────────────────────────┘
│
┌───────────────┼───────────────┐
▼ ▼ ▼
┌─────────┐ ┌─────────┐ ┌─────────┐
│ Scholar │ │ Web │ │ Smart │
└─────────┘ └─────────┘ └─────────┘
│ │ │
└───────────────┼───────────────┘
▼
┌─────────────────┐
│ AIsa Explain │
│ (元分析) │
└─────────────────┘
│
▼
┌─────────────────┐
│ 置信度评分 │
│ + 综合结果 │
└─────────────────┘
核心能力
网络搜索
bash
基础网络搜索
curl -X POST https://api.aisa.one/apis/v1/scholar/search/web?query=AI+frameworks&max
numresults=10 \
-H Authorization: Bearer $AISA
APIKEY
全文搜索(含页面内容)
curl -X POST https://api.aisa.one/apis/v1/search/full?query=latest+AI+news&max
numresults=10 \
-H Authorization: Bearer $AISA
APIKEY
学术搜索
bash
搜索学术论文
curl -X POST https://api.aisa.one/apis/v1/scholar/search/scholar?query=transformer+models&max
numresults=10 \
-H Authorization: Bearer $AISA
APIKEY
带年份筛选
curl -X POST https://api.aisa.one/apis/v1/scholar/search/scholar?query=LLM&max
numresults=10&as
ylo=2024&asyhi=2025 \
-H Authorization: Bearer $AISA
APIKEY
智能搜索(网络 + 学术结合)
bash
智能混合搜索
curl -X POST https://api.aisa.one/apis/v1/scholar/search/smart?query=machine+learning+optimization&max
numresults=10 \
-H Authorization: Bearer $AISA
APIKEY
Tavily 集成(高级)
bash
Tavily 搜索
curl -X POST https://api.aisa.one/apis/v1/tavily/search \
-H Authorization: Bearer $AISA
APIKEY \
-H Content-Type: application/json \
-d {query:latest AI developments}
从 URL 提取内容
curl -X POST https://api.aisa.one/apis/v1/tavily/extract \
-H Authorization: Bearer $AISA
APIKEY \
-H Content-Type: application/json \
-d {urls:[https://example.com/article]}
爬取网页
curl -X POST https://api.aisa.one/apis/v1/tavily/crawl \
-H Authorization: Bearer $AISA
APIKEY \
-H Content-Type: application/json \
-d {url:https://example.com,max_depth:2}
站点地图
curl -X POST https://api.aisa.one/apis/v1/tavily/map \
-H Authorization: Bearer $AISA
APIKEY \
-H Content-Type: application/json \
-d {url:https://example.com}
解释搜索结果(元分析)
bash
生成带置信度评分的解释
curl -X POST https://api.aisa.one/apis/v1/scholar/explain \
-H Authorization: Bearer $AISA
APIKEY \
-H Content-Type: application/json \
-d {results:[...],language:en,format:summary}
📊 置信度评分引擎
与标准 RAG 系统不同,OpenClaw Search 评估可信度和共识:
评分标准
| 因素 | 权重 | 描述 |
|---|
| 来源质量 | 40% | 学术 > 智能/网络 > 外部 |
| 一致性分析 |
35% | 跨来源共识检查 |
|
时效性 | 15% | 较新来源权重更高 |
|
相关性 | 10% | 查询与结果的语义匹配 |
评分解读
| 分数 | 置信度级别 | 含义 |
|---|
| 90-100 | 非常高 | 学术和网络来源高度一致 |
| 70-89 |
高 | 良好的一致性,可靠的来源 |
| 50-69 | 中等 | 信号混杂,需独立验证 |
| 30-49 | 低 | 来源冲突,谨慎使用 |
| 0-29 | 非常低 | 数据不足或相互矛盾 |
Python 客户端
bash
网络搜索
python3 {baseDir}/scripts/search_client.py web --query latest AI news --count 10
学术搜索
python3 {baseDir}/scripts/search_client.py scholar --query transformer architecture --count 10
python3 {baseDir}/scripts/search_client.py scholar --query LLM --year-from 2024 --year-to 2025
智能搜索(网络 + 学术)
python3 {baseDir}/scripts/search_client.py smart --query autonomous agents --count 10
全文搜索
python3 {baseDir}/scripts/search_client.py full --query AI startup funding
Tavily 操作
python3 {baseDir}/scripts/search_client.py tavily-search --query AI developments
python3 {baseDir}/scripts/search_client.py tavily-extract --urls https://example.com/article
多源搜索带置信度评分
python3 {baseDir}/scripts/search_client.py verity --query Is quantum computing ready for enterprise?
API 端点参考
| 端点 | 方法 | 描述 |
|---|
| /scholar/search/web | POST | 带结构化结果的网络搜索 |
| /scholar/search/scholar |
POST | 学术论文搜索 |
| /scholar/search/smart | POST | 智能混合搜索 |
| /scholar/explain | POST | 生成结果解释 |
| /search/full | POST | 带内容的全文搜索 |
| /search/smart | POST | 智能网络搜索 |
| /tavily/search | POST | Tavily 搜索集成 |
| /tavily/extract | POST | 从 URL 提取内容 |
| /tavily/crawl | POST | 爬取网页 |
| /tavily/map | POST | 生成站点地图 |
搜索参数
| 参数 | 类型 | 描述 |
|---|
| query | string | 搜索查询(必填) |
| maxnumresults |
integer | 最大结果数(1-100,默认 10) |
| as_ylo | integer |