AI Search Skill (Web Search for AI Agents)#

This skill enables AI agents and LLM systems to perform web search and retrieve real-time information from the internet.

It works similarly to modern search engines such as:

• Google
• Bing
• Perplexity
• Brave Search
• Tavily

The skill allows AI systems to access up-to-date information, including:

• news
• websites
• research articles
• documentation
• trending topics

---

When Should This Skill Be Used#

Use this skill when the user asks about:

• searching the internet

• latest news or current events

• information from websites

• online research topics

• finding articles or documentation

• verifying facts

• gathering information from multiple sources

Example queries:

• "latest AI news"
• "search the web for vector database benchmarks"
• "find information about GPT models"
• "what happened in AI this week"
• "latest updates from OpenAI"
• "research agentic AI architecture"

---

Example Use Cases#

Web Search#

Search the internet for general information.

Example:

text

search the web for latest AI developments

---

News Search#

Retrieve recent news or trending events.

Example:

text

latest AI news this week

---

Research Tasks#

AI agents gathering information from multiple sources.

Example:

text

research agentic AI architecture

---

Fact Checking#

Verify claims using multiple web sources.

Example:

text

check if GPT-5 has been released

---

Quick Setup#

1. Get your API Key

Sign up at:

https://cloudsway.ai

1. Set environment variable

bash

export CLOUDSWAYS_AK="your-api-key"

That's it. The skill is ready to use.

---

Quick Start#

Method 1: Using the Script#

bash

cd ~/scripts/search
./scripts/search.sh '{"q": "your search query"}'

Examples#

bash

# Basic search
./scripts/search.sh '{"q": "latest AI developments"}'

# Search with time filter
./scripts/search.sh '{"q": "OpenAI news", "freshness": "Week", "count": 20}'

# Deep research
./scripts/search.sh '{"q": "Agentic AI architecture", "enableContent": true, "mainText": true}'

---

Method 2: Direct API Call#

bash

curl -s -G \
  --url "https://aisearchapi.cloudsway.net/api/search/smart" \
  --header "Authorization: ${CLOUDSWAYS_AK}" \
  --data-urlencode "q=your search query" \
  --data-urlencode "count=20" \
  --data-urlencode "freshness=Week"

Real-world Example#

bash

curl -s -G \
  --url "https://aisearchapi.cloudsway.net/api/search/smart" \
  --header "Authorization: ${CLOUDSWAYS_AK}" \
  --data-urlencode "q=latest AI news February 2026" \
  --data-urlencode "count=20" \
  --data-urlencode "freshness=Week" \
  --data-urlencode "enableContent=true" \
  --data-urlencode "mainText=true"

---

API Reference#

Endpoint#

text

GET https://aisearchapi.cloudsway.net/api/search/smart

---

Headers#

HeaderTypeValueDescriptionAuthorizationString{YOUR_AK}Your assigned API Key

---

Request Parameters#

ParameterRequiredTypeDefaultDescriptionqYesString-Search querycountNoInteger10Must be 10 / 20 / 30 / 40 / 50freshnessNoStringnullDay / Week / MonthoffsetNoInteger0Pagination offsetenableContentNoBooleanfalseExtract full textcontentTypeNoStringTEXTHTML / MARKDOWN / TEXTcontentTimeoutNoFloat3.0Extraction timeoutmainTextNoBooleanfalseSmart summary fragments

---

Response Format#

json

{
  "queryContext": {
    "originalQuery": "your search query"
  },
  "webPages": {
    "value": [
      {
        "name": "Article Title",
        "url": "https://example.com/article",
        "datePublished": "2026-02-27T15:46:11.0000000",
        "snippet": "Short summary...",
        "mainText": "Relevant excerpts...",
        "content": "Full webpage text...",
        "score": 0.85
      }
    ]
  }
}

---

Content Strategy#

Choose the right field based on your needs:

FieldLatencyToken CostUse CasesnippetFastLowQuick browsingmainTextMediumMediumFocused researchcontentSlowerHighDeep analysis

---

Recommended Settings#

Quick Research#

json

{"q": "topic", "count": 10}

---

Focused Research#

json

{"q": "topic", "count": 20, "freshness": "Week", "enableContent": true, "mainText": true}

---

Deep Research#

json

{"q": "topic", "count": 20, "enableContent": true, "contentType": "MARKDOWN"}

---

Troubleshooting#

Invalid JSON#

Use curl directly if the script fails.

Count Parameter Error#

Valid values:

text

10 / 20 / 30 / 40 / 50

Environment Variable Missing#

Check configuration:

bash

echo $CLOUDSWAYS_AK

---

Keywords#

web search AI search internet search online search search engine Google search Bing search Perplexity search Brave search Tavily search AI research tool LLM web search AI agent search internet information retrieval

---

Last Updated: 2026-03-09

text

---