trawlia-mcp
Trawlia as an MCP server: web search and page extraction, as two tools any MCP client can call.
Get a free key at trawlia.co/signup - 100 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_..." }
}
}
}
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. |
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. |