Web search

Search the web and get back result titles, URLs and descriptions — optionally with each result page's content scraped to markdown in the same call. The orchestrator exposes this via the webSearch step, backed by a self-hosted SearXNG instance, with page content rendered by a headless-browser scraper when requested. Use it to ground LLM answers in live web data, verify facts, or feed pages into downstream steps like webScrape.

To let a chat model call web search itself mid-completion, use the civitai:web_search tool on the chat completion step instead.

The request shape

Every request is a single webSearch step on SubmitWorkflow:

{
  "$type": "webSearch",
  "input": {
    "query": "civitai stable diffusion models",
    "limit": 3
  }
}

Operations

Search

POST https://orchestration.civitai.com/v2/consumer/workflows?wait=30
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "steps": [{
    "$type": "webSearch",
    "input": {
      "query": "civitai stable diffusion models",
      "limit": 3
    }
  }]
}

POST/v2/consumer/workflows

Set your Civitai API token via the Token button in the navbar to enable Try It.

Request body — edit to customize (e.g. swap the image URL or prompt)

{ "steps": [ { "$type": "webSearch", "input": { "query": "civitai stable diffusion models", "limit": 3 } } ] }

Valid JSON

Search with scraped content

Add scrapeFormats to also fetch each result page and return its content inline as markdown:

POST https://orchestration.civitai.com/v2/consumer/workflows?wait=60
Authorization: Bearer <your-token>
Content-Type: application/json

{
  "steps": [{
    "$type": "webSearch",
    "input": {
      "query": "civitai stable diffusion models",
      "limit": 3,
      "scrapeFormats": ["markdown"]
    }
  }]
}

POST/v2/consumer/workflows

Set your Civitai API token via the Token button in the navbar to enable Try It.

Request body — edit to customize (e.g. swap the image URL or prompt)

{ "steps": [ { "$type": "webSearch", "input": { "query": "civitai stable diffusion models", "limit": 3, "scrapeFormats": [ "markdown" ] } } ] }

Valid JSON

Parameters

Field	Default	Allowed	Notes
query	— ✅	string	The search query.
limit	5	1–10	Maximum number of results.
scrapeFormats	null	["markdown"]	When set, each result page is scraped and its content returned inline. Slower — every result costs a page render.

Reading the result

{
  "status": "succeeded",
  "steps": [{
    "name": "0",
    "$type": "webSearch",
    "status": "succeeded",
    "output": {
      "results": [
        {
          "url": "https://civitai.com/models",
          "title": "AI Models | Civitai",
          "description": "Browse thousands of free Stable Diffusion models...",
          "markdown": null
        }
      ]
    }
  }]
}

results is ordered by relevance. markdown is only populated when scrapeFormats was set. Results can be fed into other steps with $ref — e.g. a repeat step scraping every hit via { "$ref": "search", "path": "output.results", "as": "result" }.

Identical queries are cached for a while, so repeating the same search is fast and free of extra crawling.

Runtime

Variant	Per-call wall time	wait recommendation
Search only	1–5 s	wait=30
With scrapeFormats	5–30 s	wait=60

For anything past the 100 s request timeout, use webhooks — see Results & webhooks.

Cost

Billed in Buzz on the workflow's transactions. Use whatif=true for an exact preview; see Payments (Buzz).

Operation	Buzz
Search (with or without scrapeFormats)	1

Troubleshooting

Symptom	Likely cause	Fix
Empty results	Query too narrow, or the search backend returned nothing	Rephrase the query; try a higher limit.
markdown is null on some results with scrapeFormats set	The result page blocked scraping or timed out	Expected for heavily bot-protected sites — use the description, or try webScrape on a different URL.
Step stays queued	No web-search workers available	Capacity issue — retry later or contact us.

Related

• SubmitWorkflow — operation used by every example here
• GetWorkflow — for polling
• Web scrape — fetch a single known URL
• Chat completion — let the model search the web itself via civitai:web_search
• Full parameter catalog: the WebSearchInput schema in the API reference