nodecode-anysearch

AnySearch: web search and page extract through AnySearch's REST API, with the MCP extract fallback beside it. A Nodecode rewrite of vollegrewar/hermes-plugin-anysearch by vollegrewar.

The model gets two tools: anysearch_search, which returns titles, links and an abstract per result, and anysearch_extract, which fetches each URL as text. Both talk to AnySearch's own REST API. /anysearch QUERY is the operator's own search, and (help :anysearch) is the model's page.

Get it

Tell the agent "add the anysearch cell", or run nodecode add nc://anysearch.

Set it up

Nothing is required: AnySearch answers without a key, at a lower quota.

A free key from anysearch.com raises the quota to 1,000 requests a day. Put it in the environment:

echo 'ANYSEARCH_API_KEY=…' >> ~/.nodecode/.env

or in the cell's section, which /settings writes:

{
  "anysearch": {
    "api_key_env": "ANYSEARCH_API_KEY",   // or "api_key" written inline
    "max_results": 5,
    "max_urls": 5
  }
}

With an Authorization header present but the key invalid, the API answers 401 or 403 rather than falling back to the anonymous tier, and the tool says so in its answer.

Setting	Default	What it does
`api_key` / `api_key_env`	none	the key, inline or by environment variable (`ANYSEARCH_API_KEY` is read either way)
`base_url`	`https://api.anysearch.com`	the API root, for a proxy of your own
`max_results`	`5`	results one search asks for, at most 10
`max_urls`	`5`	URLs one extract fetches, one request each
`timeout_seconds`	`30`	one search's ceiling
`extract_timeout_seconds`	`60`	one page fetch's ceiling
`mcp_fallback`	`true`	ask the `/mcp` endpoint when `/v1/extract` answers no content or fails

Use it

anysearch_search — {query, limit?}. Answers one line per result: its title, its URL and an abstract clipped to 400 characters. limit is clamped to 1..10; unset, the section's max_results decides.
anysearch_extract — {urls}. One request per URL, in the order given, each page's text under its title. A URL that fails says so on its own line and the rest still answer, as the original's per-URL error objects do.
/anysearch QUERY — the same search, for the operator. With no argument it says where the cell points and whether it has a key.

What is in it

File	What it is
`cell.lisp`	the section, the endpoints and the MCP fallback, the two tools, `/anysearch` and `(help :anysearch)`
`test/cell-test.lisp`	lifecycle, both tools, the fallback, `/anysearch` and the refusals, every request a stubbed `nlk:http`
`config.example.jsonc`	the section, commented

Why two tools, and not a search backend

In Hermes the plugin registered itself as three web providers behind the standard web_search and web_extract tools. Nodecode's nodecode-websearch cell keeps its engines in a fixed list a cell cannot add one to, so AnySearch is two verbs of its own here. The endpoints, the request shapes and the answers are the original's; only the seam differs.

The plugin's own MCP extract fallback came with it: when /v1/extract answers no content or fails, the same URL is asked of /mcp, whose content[].text is a JSON string that is parsed once more. The two chains the plugin also shipped, Exa → AnySearch and Firecrawl → AnySearch, were left out: they rank another provider first, and that order belongs to the caller, not to this cell.

Test

From the cells repository's root, offline:

.github/test-cell.sh tools/nodecode-anysearch

nc://anysearch