About Trawlia
Trawlia is an MCP server for agents that need live web or news search and readable page extraction. It returns ranked search results and cleaned page text through two tools.
Key features
- 01Search the live web
- 02Search news with publication dates
- 03Filter results by domain and time
- 04Extract readable page content
- 05Report source coverage and read failures
Use cases
- 01Find current information on a topic
- 02Search news within a time range
- 03Extract text from a list of URLs
MCP support
Exposes `trawlia_search` and `trawlia_extract` as tools for MCP clients. A Trawlia API key is required.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y trawlia-mcpTRAWLIA_API_KEYSecretA Trawlia API key. Free at https://trawlia.co/signup, 100 credits a month, no card.TRAWLIA_BASE_URLOverride the API origin. Defaults to https://api.trawlia.co; only needed to point at a local server.Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
2Saved tool list · updated 2w ago
| Parameter | Type | Required | Description |
|---|---|---|---|
| query | string | Yes | The question, in natural language rather than keywords. |
| topic | enum | No | 'news' for recent events; it returns publication dates that 'general' usually lacks.general | news |
| time_range | enum | No | Window back from now. Drops results with no publication date, so it filters hard on topic 'general'.day | week | month | year |
| max_results | integer | No | Fewer is usually better; these results go into your context. |
| search_depth | enum | No | 'basic' costs 1 credit and returns snippets. 'advanced' costs 2 and returns full page text. Prefer basic.basic | advanced |
| exclude_domains | array | No | |
| include_domains | array | No | Restrict to these hosts. Subdomains are included. |
Details
From the project READMEtrawlia-mcp
Trawlia as an MCP server: news search, web search and page extraction, as two tools any MCP client can call.
Get a free key at trawlia.co/signup - 1,000 credits a month, no card.
Claude Code
claude mcp add trawlia --env TRAWLIA_API_KEY=twl_... -- npx -y trawlia-mcp
Claude Desktop, Cursor, and anything else reading a JSON config
{
"mcpServers": {
"trawlia": {
"command": "npx",
"args": ["-y", "trawlia-mcp"],
"env": { "TRAWLIA_API_KEY": "twl_..." }
}
}
}
The tools
trawlia_search
Ranked, deduplicated results from the live web.
| Argument | Default | Notes |
|---|---|---|
query | required | Natural language, not keywords. |
search_depth | basic | basic is 1 credit and returns snippets. advanced is 2 and returns full page text. |
topic | general | news reaches news indexes and returns publication dates. |
max_results | all | Top N of the ranked list. Left out, every result found is returned, typically 20 to 60. These go into your context, so set it when the task is narrow. |
include_domains / exclude_domains | - | Host filters. Subdomains are included. |
time_range | - | day, week, month, year. |
trawlia_extract
Fetches URLs you already have and returns their readable content, with navigation, cookie banners and newsletter modals removed. 1 credit per 5 URLs, so pass the whole list at once.
What the model sees
Results come back as readable text rather than raw JSON, and carry two things a plain response would lose:
- When a page could not be read, the result says so and why, instead of presenting a snippet as though it were the article.
- A coverage line when a source was unavailable, so the model can weigh a thin result set properly rather than reporting it as a confident nothing.
Errors are readable sentences that say whether retrying is worth it. Running out of credits says so and says not to retry; a rate limit says how long to wait.
Costs
| Operation | Credits |
|---|---|
| Basic search | 1 |
| Advanced search | 2 |
| Extraction | 1 per 5 URLs |
Charged before the work and refunded on hard failure. Full price list at trawlia.co/pricing.
Not available
Search and extraction, and nothing else. There is no crawler and no long-running job to poll. If your agent needs to walk a whole site, this is the wrong server for it.
Checking it works
TRAWLIA_API_KEY=twl_... node smoke.mjs
Spends a few credits against the real API and prints what a client sees: the tool list, a rendered search, a rejected parameter, and an extraction. Not in CI, because it costs money and depends on the live web.
Environment
| Variable | Required | Notes |
|---|---|---|
TRAWLIA_API_KEY | yes | A search-scoped key is enough and is what you should deploy. |
TRAWLIA_BASE_URL | no | Defaults to https://api.trawlia.co. For pointing at a local server. |
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.