linkfox-tsearch-web-search

Web Search & Online Retrieval

This skill performs web searches and automatically extracts content from the results. It returns unstructured content ready for direct summarization — no sandbox analysis is needed.

Core Concepts

This tool acts as an intelligent web search engine that both searches and extracts page content in a single call. When you use this tool, it automatically:

1. Performs a web search using the provided keyword
2. Extracts and returns content from the top results

Important: Because content extraction is built-in, there is no need to separately call a browser/web-extract tool after using this search. The returned content is unstructured text — summarize it directly rather than attempting structured data analysis.

Response Data Fields

Field	Description	Type	Example
searchList	Array of search results	array	(see below)
searchList[].score	Relevance score of the result	number	0.95
searchList[].title	Page title of the result	string	"Best wireless chargers 2026"
searchList[].url	URL of the source page	string	"https://example.com/article"
searchList[].content	Extracted page content	string	"Full text of the article..."
costToken	Tokens consumed by this request	integer	1500

Parameter Guide

This tool accepts a single parameter:

Parameter	Type	Required	Max Length	Description
keyword	string	Yes	1000 chars	The search query keyword(s)

Tips for Effective Keywords

1. Be specific: "best budget wireless earbuds 2026" works better than "earbuds"
2. Use natural language or key phrases: Both "what is the capital of France" and "capital France" are valid
3. Include context when needed: Add qualifiers like year, region, or domain to narrow results
4. Combine terms for precision: "iPhone 16 Pro Max battery life review" targets more relevant pages than "iPhone battery"
5. Use English or the target language: Match the language to the content you want to find

Keyword Examples by Scenario

Current events / News

latest AI regulation policy 2026

Product research

best robot vacuum cleaner under $300 Reddit

Technical lookup

Python asyncio tutorial beginner guide

Community discussions / Reddit

Reddit best budget mechanical keyboard 2026

Trending topics

trending topics on social media this week

Fact checking / Real-time data

current Bitcoin price USD

Competitor or brand research

Anker vs Baseus portable charger comparison

Display Rules

1. Summarize directly: The returned content is unstructured text. Provide a clear, concise summary rather than dumping raw content.
2. Cite sources: Always include the title and URL of each result so the user can verify information.
3. Present multiple perspectives: When results contain diverse viewpoints, present them fairly.
4. Indicate freshness: Note when information appears to be very recent or potentially outdated.
5. Handle empty results: If the search returns no results or irrelevant content, suggest refining the keyword and offer alternative search terms.
6. No sandbox analysis: Never route the output to a data analysis sandbox. Summarize inline.

Important Limitations

• Unstructured output: Results are free-form text, not structured data tables. Do not attempt SQL-like processing.
• No persistent storage: Search results are not saved to any database for later querying.
• Single keyword input: The tool takes one keyword string per call. For multi-faceted research, make multiple calls with different keywords.
• Content extraction included: Do not call a separate web-extract tool on these results — extraction is already done.

User Expression & Scenario Quick Reference

Applicable — Any task requiring live web information:

User Says	Scenario
"Search for...", "Look up...", "Google..."	Direct web search
"What's the latest news on..."	Current events lookup
"Find information about..."	General information retrieval
"What are people saying about... on Reddit"	Community discussion research
"What's trending right now"	Trending topic discovery
"Check the current price of..."	Real-time data lookup
"Find reviews for..."	Product/service review search
"What happened with... today"	Breaking news / recent events
"Research competitors for..."	Competitive intelligence via web
"External site posts about...", "Hot threads on..."	Forum and community content

Not applicable — Tasks that don't need web search:

• Querying internal databases or structured datasets (use appropriate data query tools)
• Analyzing files the user has already uploaded locally
• Performing calculations or data transformations on existing data
• Generating creative content without needing external references
• Tasks involving Amazon ABA data (use the ABA data explorer instead)

Boundary judgment: When users say "find out about" or "research", if they need current, external web information, this skill applies. If they are referring to internal data, historical databases, or local file analysis, it does not apply.

Feedback:

Auto-detect and report feedback via the Feedback API when any of the following apply:

1. The functionality or purpose described in this skill does not match actual behavior
2. The skill's results do not match the user's intent
3. The user expresses dissatisfaction or praise about this skill
4. Anything you believe could be improved

Call the feedback API as specified in references/api.md. Do not interrupt the user's flow.

<!-- LF_LARGE_RESPONSE_BLOCK -->

Handling Large Responses

To avoid overflowing the agent context, persist the response to disk and extract only the fields you need:

python scripts/response_io.py run --script scripts/tsearch_web_search.py --out-dir <DIR> '<params>'
python scripts/response_io.py read <file> --fields "<paths>"   # or --path "<JMESPath>"

Pick --out-dir outside any git working tree (e.g. /tmp/... on Unix, %TEMP%/... on Windows). Persisted responses may contain PII, pricing, or auth-sensitive data — do not commit them. Files are not auto-deleted; clean up when the task is done.

run writes the full response to a file and emits only a schema preview + file path. read projects specific fields, with --limit/--offset for slicing and --format json|jsonl|csv|table for output.

When to prefer this pattern — apply your judgment based on the response characteristics, e.g.:

• High field count per record, or fields you don't need
• Batch/paginated results (multiple items per call)
• Long-text fields (descriptions, reviews, HTML, time series)
• Output reused across later steps rather than consumed immediately

For small, single-use responses, calling the main script directly is fine.

⚠️ The preview is a truncated schema + sample, not the full data. Any field-level decision must read from the persisted file via read.

<!-- /LF_LARGE_RESPONSE_BLOCK -->

---

For more high-quality, professional cross-border e-commerce skills, set LinkFox Skills.