Skip to main content
Glama
Sign In

Tulina

by ai.tulina.mcp
OverviewDiscussionsAdmin

serper_search

Read-only

Google search via Serper — one vertical per kind.

kind:

  • "web" (default): organic results. Accepts site_filter (e.g. "linkedin.com/in"), autocorrect, location, tbs.

  • "news": Google News — useful to monitor a target's signals (press releases, hires, fundraises). Accepts tbs. ⚠️ This vertical does NOT honour boolean syntax: OR, quoted phrases or site: come back as an EMPTY news list, not as "no news". Split an A OR B query into one call per term, plain words only.

  • "images" / "videos": return an images / videos array (title, link, source, dimensions or duration). Accept tbs.

  • "places": Google Local — the businesses for a query. Excellent for local B2B prospecting: title, address, phone, website, rating, review count and cid (to pass to serper_reviews). Accepts location.

  • "shopping": title, price, merchant, rating, delivery. Accepts location.

  • "scholar": academic publications (title, journal, year, citations, pdf).

  • "patents": patents (title, inventor, applicant, number, dates).

  • "autocomplete": Google suggestions for query — to widen a lexical field or find keyword ideas. Ignores pagination.

Under a project with excluded_url_prefixes, the matching results are dropped and counted.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "query": "<query>"
    },
    "name": "serper_search"
  }
}

Input Schema

NameRequiredDescriptionDefault
numNonumber of results (max 100). Ignored by "autocomplete".
tbsNoGoogle time filter — "qdr:d" (24 h), "qdr:w" (7 d), "qdr:m"… Accepted by web / news / images / videos.
_orgNoOrg (id) de CET APPEL. `oto_list_orgs()`.
fullNoreturns the ENTIRE Google response. By default (recommended) the return is narrowed to what a sweep reads — title, link, snippet — and drops knowledge graph, people-also-ask, related searches and per-result sitelinks (~a third of the payload, never read). Only pass `full=True` if you specifically want one of these sections. Accepted by web / news.
kindNothe vertical (default "web"): web | news | images | videos | places | shopping | scholar | patents | autocomplete.web
pageNoresults page (1-based). Ignored by "autocomplete".
queryYesthe query.
_groupNoÉquipe (id) de CET APPEL, sous son org parente.
fieldsNokeep ONLY these keys on each result (e.g. ["title","link", "snippet"]). The envelope (credits, pagination) is kept in all cases. Accepted by web / news.
_run_idNoOBLIGATOIRE à chaque appel dès `run_start` ou `data_claim_next` : sans lui, l'écriture sur une ligne réservée est refusée. Hérité seulement si le serveur tient un run actif — n'y compte pas.
countryNocountry code (default "fr").fr
_projectNoProjet (id) de CET APPEL — jeton PRIMAIRE. `oto_project(op='list')`.
languageNolanguage code (default "fr").fr
locationNogeographic bias (e.g. "Paris, France"). Accepted by web / places / shopping.
_instanceNoInstance de connecteur PRÉCISE (un credential exact). `oto_instance(op='list')`.
autocorrectNokind="web" — toggles Google spelling correction.
site_filterNokind="web" — restrict to a domain (e.g. "linkedin.com/in").

Schema Changelog

Changes observed during successful MCP inspections.

No schema history has been recorded yet.

TDQS

A4.5/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only say readOnlyHint=true; the description adds non-obvious behavior an agent could not infer — that the news vertical rejects boolean syntax and returns an EMPTY news list rather than 'no news', with the remediation of one call per plain term. It also discloses that project-level excluded_url_prefixes silently drops matching results (but counts them) and that autocomplete ignores pagination, which are exactly the failure modes that change query strategy.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded one-liner followed by a bolded per-vertical list with defaults and caveats inline; every bullet carries information an agent needs to choose and shape a call. Minor overhead comes from restating the kind enum values and accepted parameters that the schema already enumerates, but the density is justified for nine verticals.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 17-parameter tool with no output schema, the description compensates well by naming the fields each vertical returns (title/link/snippet, cid, price/merchant, citations/pdf) and by explaining the reduced-payload default and the project exclusion filter. It leaves the envelope/credits behavior and the cross-cutting _project/_org/_run_id context params unaddressed, which is the only gap relative to the tool's complexity.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, so baseline would be 3, but the description earns above baseline by mapping which parameters each vertical actually accepts (site_filter/autocorrect/location/tbs for web, location for places/shopping, tbs only for news/images/videos) and by stating per-vertical output fields. The description does not touch the underscore-prefixed context params (_project, _org, _run_id, _instance), leaving those entirely to the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The opening line names a specific verb and resource (Google search via Serper) and the exact control that selects behavior across nine verticals. Each vertical is then described by what it returns, so an agent can distinguish it from serper_scrape, serper_lens, serper_maps_sample and serper_reviews without opening any schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Guidance is concrete and situational: news is flagged for monitoring a target's signals, places for local B2B prospecting, autocomplete for widening a lexical field, and the cid output is explicitly routed to serper_reviews. What's missing is any when-not guidance or routing toward the other Serper siblings (lens, scrape, maps_sample), so an agent has no stated cue for when to prefer those over this tool.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Tulina toolsInspect
Try in Browser
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources