Skip to main content
Glama

Visibility Genies

Run a site audit

run_site_audit

Starts a site audit. It crawls the site (same origin, following robots.txt), checks each page for SEO problems such as broken links, missing or duplicate titles and descriptions, redirect chains, orphan pages, canonical mistakes and pages blocked by robots.txt, and can run Lighthouse on a sample of pages. The crawl is free. Lighthouse is paid (up to 10 pages on mobile and desktop, up to 20 tests, up to $0.20) and is billed while the audit runs. The audit works in the background: poll get_audit_status, then read get_audit_issues, and get_audit_pages for Lighthouse scores. When a site rate limits the crawler it backs off and retries, and pages it still cannot read are marked blocked or rate-limited instead of being guessed at. Every audit crawls at one gentle pace, about a page a second, and crawlSpeed is accepted but ignored. The crawler slows down by itself; if the site keeps refusing, allow our crawler by name and address (the bot page lists both). The crawler calls itself VisibilityGeniesBot unless crawlerIdentity says otherwise: desktop or mobile send a browser user agent that keeps that token, while chrome_windows, firefox_windows, safari_macos, chrome_android and safari_iphone send an ordinary browser's headers without it, and none imitates a search engine. Pages are read as HTML, with no JavaScript, unless renderJavaScript is true. The project Owner may let audits of the project's own site ignore robots.txt, and those audits report a robots-txt-ignored finding. Each call may spend up to the workspace's agent spend limit ($2.00 unless changed in Settings) without asking: it holds at most its estimate plus a margin while it runs and is charged what it bought, never more. A call whose maximum is above the limit is refused before anything is bought, with its estimate and maximum: ask the user, then call again with approvedMaxCostUsd set to the amount they approved.

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": {
      "projectId": "<projectId>",
      "url": "<url>"
    },
    "name": "run_site_audit"
  }
}

Input Schema

NameRequiredDescriptionDefault
urlYesStart URL to crawl.
maxPagesNoPage budget for the crawl (default 50).
crawlModeNosearch_engine (default) follows links outward from the start page the way a search engine discovers pages. full_inventory also seeds every sitemap URL so orphan pages show up.search_engine
projectIdYesRequired. The Visibility Genies project this call works in and is billed to. Call list_projects to find one.
crawlSpeedNoIgnored. Every audit crawls at one gentle pace, about a page a second.
ignoreParamsNoQuery parameters to strip before comparing URLs, so ?page=1 and ?page=2 count as one page (up to 50).
disallowPathsNoPath prefixes to leave out, such as /tag or /search (up to 50). A prefix excludes everything beneath it.
runLighthouseNoAlso run Lighthouse. This is paid: up to 10 pages on mobile and desktop, up to 20 tests, up to $0.20. Defaults to false and adds several minutes. Set it only when the user wants performance detail from a lab test (LCP, CLS, Total Blocking Time), which is not real visitor data.
crawlerIdentityNoWhat the crawler calls itself. seogeniesbot is the default and sends the VisibilityGeniesBot user agent. desktop and mobile send a Chrome user agent that still names VisibilityGeniesBot. chrome_windows, firefox_windows, safari_macos, chrome_android and safari_iphone copy an ordinary visitor's request headers with no VisibilityGeniesBot token and never use the project's allow key. No choice pretends to be a search engine crawler, and all of them obey the robots.txt rules for VisibilityGeniesBot and for its former name, SEOGeniesBot. Browser identities only crawl the project's own site (its domain or the www form); any other start URL is refused. renderJavaScript works with chrome_windows and chrome_android, not with Firefox or Safari. When the Owner turned on the faster crawl for the project's own site, an audit of that site crawls as seogeniesbot whatever this says.seogeniesbot
renderJavaScriptNoLoad each page in a browser before reading it, for sites that build their pages with JavaScript. Defaults to false. It is slower, covers up to 500 pages per audit, and uses a hosted browser the platform pays for. Where it is not available the audit is refused and says why.
approvedMaxCostUsdNoThe most, in USD, the user approved for this call after seeing its estimate and maximum. Only needed after a call was refused for costing more than the workspace's agent spend limit; never set it without asking the user.

Output Schema

NameRequiredDescriptionDefault
costNo
metaNo
auditIdNo
savedResultIdNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.7/5.0
Behavior5/5

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

Annotations only declare non-read-only, open-world and non-destructive; the description adds the costly, hard-to-infer behaviors: crawl is free but Lighthouse is paid and billed while running, the audit is asynchronous, the crawler backs off on rate limits and marks unreadable pages as blocked/rate-limited rather than guessing, and owner-enabled robots.txt bypass surfaces as a robots-txt-ignored finding. The spend-limit mechanics — provisional hold, refusal before purchase, and approvedMaxCostUsd retry — are exactly the kind of context an agent cannot get from structured fields.

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

Conciseness3/5

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

It is front-loaded correctly with the purpose and the polling flow, but it runs to a dense, repetitive wall of prose — the ignored crawlSpeed and the fixed gentle crawl pace are stated twice, and the crawler-identity block largely duplicates the schema's own long enum description. The content earns its place individually, but the lack of structure makes it slower to parse than it needs to be.

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

Completeness5/5

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

For an 11-parameter, paid, asynchronous, open-world tool, the description covers what an agent needs: cost model, refusal/approval loop, async polling, crawler identification constraints, rendering limits, and rate-limit fallback. Return values need not be explained because an output schema exists and the description already points to the tools that read results.

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 the baseline is 3, but the description adds decision-level meaning: crawlSpeed is accepted but ignored, identity choices affect JavaScript rendering and are restricted to the project's own site, and approvedMaxCostUsd is only legitimate after a refusal and only with user approval. It mostly restates enumeration semantics already present in the schema descriptions, which keeps it below a 5.

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 sentence names a specific verb and resource, and the rest enumerates exactly what the crawl checks (broken links, titles, descriptions, redirect chains, orphan pages, canonicals, robots.txt blocks) plus optional Lighthouse. It is unmistakably distinct from siblings like get_audit_status or check_page_seo, which only read results of an audit started here.

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

Usage Guidelines5/5

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

It explicitly routes the agent through the follow-up flow: start here, then poll get_audit_status, then read get_audit_issues and get_audit_pages for Lighthouse scores. It also states conditional guidance for the paid path ('set it only when the user wants performance detail from a lab test') and the exact escalation path when a call is refused for exceeding the spend limit.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources