Tavily Web Search¶
Queries the Tavily web search service via Salt Agents and returns relevant results. Optionally includes a concise AI-generated answer alongside a JSON string of retrieved documents. Supports search depth, topic scoping (general/news), domain inclusion/exclusion, and toggles for raw content.
Usage¶
Use this node to fetch up-to-date web information for research, summaries, or grounding downstream reasoning. Typical workflow: provide a clear query, choose depth (basic/advanced), set result limits, optionally scope to news with a recent-days window, and refine with include/exclude domain lists. Feed the 'documents' output into parsers, summarizers, or retrieval chains; use 'answer' for quick overviews.
Inputs¶
| Field | Required | Type | Description | Example |
|---|---|---|---|---|
| query | True | STRING | Your search query. Provide a clear, specific request describing the information you need. | What are the latest developments in quantum error correction? |
| search_depth | True | STRING | The depth of the search. Allowed values: basic (1 credit), advanced (2 credits). Advanced may retrieve broader/deeper coverage. | advanced |
| max_results | True | INT | Maximum number of search results to return. Minimum is 1. | 5 |
| topic | True | STRING | Category of the search. Allowed values: general, news. Choose 'news' for recent coverage focused on current events. | news |
| days | True | INT | Only applies when topic is 'news'. Number of days back from today to include in results. | 3 |
| include_answer | True | BOOLEAN | If true, includes a short synthesized answer to the original query. | True |
| include_raw_content | True | BOOLEAN | If true, includes raw text content of search results. If false, only the most query-relevant extracted content is returned. | False |
| include_domains | True | STRING | Domains to specifically include in results. Provide one domain per line. | nature.com science.org arxiv.org |
| exclude_domains | True | STRING | Domains to exclude from results. Provide one domain per line. | wikipedia.org reddit.com |
Outputs¶
| Field | Type | Description | Example |
|---|---|---|---|
| answer | STRING | A concise answer to the query, when available and include_answer is enabled. | Quantum error correction has advanced with new bosonic codes reducing overhead and demonstrations of logical qubit lifetimes exceeding physical qubits in 2025 studies. |
| documents | STRING | A JSON-formatted string containing an array of document objects returned by the search (titles, URLs, snippets, and optionally raw content). | [ { "title": "Breakthrough in Quantum Error Correction", "url": "https://example.com/article", "snippet": "Researchers report improved logical qubit fidelity...", "content": "..." } ] |
Important Notes¶
- Query must be non-empty: An empty query will raise an error.
- News-only window: The 'days' parameter only affects searches when topic is set to 'news'.
- Answer availability: If include_answer is disabled or no answer is produced, the 'answer' output may be empty.
- Domain filters: Enter one domain per line for include/exclude lists; leave empty to not constrain by domains.
- Result format: 'documents' is a JSON string; parse it before downstream processing if you need structured access.
- Search depth costs: 'advanced' depth uses more credits than 'basic'.
Troubleshooting¶
- Empty or missing results: Increase max_results, switch to 'advanced' search_depth, broaden the query, or remove restrictive domain filters.
- No answer returned: Ensure include_answer is true and try 'advanced' depth; some queries may not yield a synthesized answer.
- Stale or irrelevant results: Use topic='news' with an appropriate 'days' value to focus on recent content, or specify include_domains.
- Errors about empty query: Provide a non-empty 'query' string.
- Network or service errors: Retry later; verify your Salt Agents service configuration and connectivity.
- Downstream parsing errors: Remember 'documents' is a JSON string; parse it before attempting to access fields.
Example Pipelines¶
