Mark

Get Site Audit

get_site_audit

Read-onlyIdempotent

Returns the latest site audit: score, issues by severity with affected pages, the prioritized fix plan, and aiCrawlerAccess: what the site's robots.txt says to each AI crawler (allowed, blocked, no rule, or unknown), not whether an edge or WAF lets it through. Each page carries render: how a crawler that runs no JavaScript reads it, as class js_heavy (needs JavaScript), static_html (pre-built HTML) or raw_html (content in the HTML, no sign it was pre-built), or null with why when its raw fetch failed, plus keyPage (homepage, pricing, compare, demo, contact and similar); a js_heavy key page is a fix candidate. Sites that serve AI bots pre-rendered HTML by user agent (dynamic rendering) still read as js_heavy, because Mark fetches with its own user agent. Four checks are recommendations that do not change the score: article_schema_incomplete (a post's Article JSON-LD lacks recommended fields such as a Person author), schema_only_after_js (JSON-LD that appears only after JavaScript runs), ai_crawler_blocked (robots.txt blocks an AI search or user-fetch bot) and hub_missing (the homepage header and footer miss one or more of these sections: features or services, pricing, compare, resources, and company; detail.missing lists them). Two site-structure checks do count: money_page_few_links (a pricing, feature, service, product or comparison page with fewer than 2 internal links; detail.inboundLinks) and standard_page_missing (no Privacy, Terms, or Contact or Support page, or the one found returns 4xx or 5xx or has under 50 words; detail.missing and detail.broken). agenticReadiness holds Google Lighthouse's experimental agentic browsing checks (Chrome 150+, may change), run through PageSpeed Insights on up to three pages: each check is pass, fail, na, informative or error (couldn't check), read as N of M passed over pass and fail only, never a score; llms.txt is Mark's own fetch by Lighthouse's rules; markdown (what Accept: text/markdown returns) and links (the homepage's Link header) are Mark's facts, not scored; a failing check is also an informational agentic_* issue. With saved identity links, the incomplete Organization schema finding points (appUrl) at the Brand page's Organization JSON-LD, which get_brand_context returns as organizationJsonLd. aiCrawler404s (primary site with PostHog connected, else empty) lists pages AI assistants can't find: paths assistants fetched 3+ times in 7 days, over 2+ days or from 2+ assistants, that answered 404 on Mark's daily check, each with hits, assistants, checkedAt and a recommendation: redirect (a 301 to url, with the matchedTerms), keep_404 (no confident match; redirecting to an unrelated page is a soft 404), no_audit (run an audit for a suggestion) or publish_file (robots.txt, a sitemap, llms.txt or /.well-known/: publish the file rather than redirect). Dismissed paths are left out. A site with no audit yet returns audit null when it has aiCrawler404s, and an error otherwise. Poll this after run_site_audit. Related: run_site_audit, check_site_readiness, get_brand_context, dismiss_ai_crawler_404.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`siteId`	Yes	Site id from list_sites.

Output Schema

TableJSON Schema

Name	Required	Description	Default
`audit`	Yes
`appUrl`	Yes
`aiCrawler404s`	Yes

Schema Changelog

Changes observed during successful MCP inspections.

Changed2 schema fields changedOct 20, 2026

addedOutput schema / properties / aiCrawler404s

Added value: +{
+  "items": {},
+  "type": "array"
+}

changedOutput schema / required

Previous value: -[
-  "appUrl",
-  "audit"
-]New value: +[
+  "aiCrawler404s",
+  "appUrl",
+  "audit"
+]

First observedOct 1, 2026

TDQS

A4/5.0

Behavior5/5

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

Annotations already cover safety (readOnly, idempotent, non-destructive, closed-world), and the description still adds substantial behavioral context: the null-audit-vs-error behavior, that dismissed paths are excluded, that recommendations do not change the score, that agenticReadiness is reported as 'N of M passed, never a score', and that dynamic rendering is misread as js_heavy. That is real disclosure 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.

Conciseness2/5

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

A single unbroken block of roughly 500 words with no headings, lists, or paragraph breaks, mixing purpose, output semantics, edge cases and sibling pointers. Much of the return-value detail (render classes, check names, agenticReadiness states) restates what the existing output schema already encodes, and the density makes the important routing sentence ('Poll this after run_site_audit') easy to miss.

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?

Given the tool's complexity, the description is more than adequate: it covers the no-audit case, the error case, scoring vs non-scoring checks, and pointers into get_brand_context and dismiss_ai_crawler_404. Nothing an agent needs in order to call this one-parameter read tool correctly is absent.

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

Parameters3/5

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

One parameter with 100% schema description coverage ('Site id from list_sites.'), so the schema already carries full meaning and the baseline is 3. The description adds no format, source, or validation guidance for siteId, and none is needed.

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

Purpose4/5

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

Opens with a specific verb and resource: 'Returns the latest site audit', then enumerates the payload (score, issues by severity, fix plan, aiCrawlerAccess). An agent immediately knows this is a read of an existing audit rather than a trigger. It lists related siblings at the end but never states the decision rule that separates it from check_site_readiness or get_brand_context, so it stops short of a 5.

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?

'Poll this after run_site_audit' gives a concrete temporal trigger, and the 'Related:' line names run_site_audit, check_site_readiness, get_brand_context and dismiss_ai_crawler_404. However, it only lists the alternatives without saying when each is preferred or excluded, and the tie-in with dismiss_ai_crawler_404 is only implied by the 'Dismissed paths are left out' sentence.

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

All Mark tools Inspect

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

Need Help?