OpenRush

Pull keyword, SERP, backlink, site-audit and AI-citation data, plus Search Console and GA, into an AI agent.

Add to Agentman Studio

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

Resources

What the OpenRush connector does

OpenRush is a pay-per-call SEO and competitor research data service. Its MCP server gives Claude, ChatGPT and any MCP-compatible agent eighteen read-only tools for keyword, SERP, backlink, site-audit and AI-citation data, plus your own connected Google Search Console and Google Analytics. Nothing writes. Every call spends prepaid credits.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

OpenRush tools (18)

  • audit_site
  • compare_backlink_gap
  • compare_keyword_coverage
  • describe_capabilities
  • discover_competitors
  • export_dataset
  • get_search_performance
  • inspect_backlinks
  • inspect_domain
  • inspect_keyword
  • inspect_page
  • inspect_search_visibility
  • inspect_serp
  • list_websites
  • research_keywords
  • get_website_analytics

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • Every call spends prepaid credits, and no tool can read the balance. Sixteen of the eighteen tools cost between 1 and 40 credits, deducted when the call starts. Nothing in the MCP surface reads a balance, quota or usage record. Do not blanket-approve this connector; approve the 20-, 22- and 40-credit calls individually.
  • Anthropic's directory under-lists the server by two tools, and both omissions are expensive read calls. Budget from OpenRush's own credits table or its live capability endpoint, not from the directory listing.
  • AI-visibility coverage is Google AI Overview only. OpenRush states plainly that inspect_ai_visibility and discover_ai_citations do not cover ChatGPT, Claude, Gemini, Perplexity or Copilot, that the metric is "your domain cited as a source" rather than how often AI discusses you, and that results are a sampled observation to be read as directional. At 20 and 22 credits these are the calls most likely to be over-read.
  • The site audit is a sample, not a crawl. audit_site reads at most 20 pages, defaults to 10, parses raw HTML only, does not execute JavaScript and does not measure Core Web Vitals. JavaScript-rendered sites will look emptier than they are.
  • Search Console reporting stops at sixteen months because that is Google's retention limit, the most recent three days are marked preliminary, and it covers only pages that received impressions. It reports no index-coverage or page-health data.
  • inspect_domain returns no link authority. OpenRush documents this twice, because agents get it wrong: the domain snapshot uses the organic search index only. Link metrics require inspect_backlinks, a separate 9-credit call.
  • Backlink tools can be switched off. OpenRush documents an operational kill switch that can remove inspect_backlinks and compare_backlink_gap from a deployment, and directs callers to its capability endpoint as the live source of truth. Both were present on 2026-08-24.
  • The vendor's own documentation contradicts itself on one price. OpenRush's tool reference states discover_competitors costs 3 credits, while its credits table, its pricing page and its live capability endpoint all say 7. Three sources to one, so 7 is almost certainly correct — but a reader taking the tool reference at face value would under-budget it.
  • No OAuth scope narrows anything. scopes_supported is an empty array and the authorization server offers identity claims only. This is benign on a read-only surface, but a token here is account-wide and spends the same balance as your dashboard and API keys.
  • We could not read tool schemas or safety annotations. The MCP endpoint returned 401 to an anonymous request, so no readOnlyHint or destructiveHint is published here. The read-only classification comes from vendor-documented request schemas corroborated by the live capability endpoint.
  • Retry behaviour on an expensive call is unresolved. OpenRush documents no idempotency key and no retry semantics for tool calls, and charges at call start. We did not test whether a retried 40-credit call is charged twice; treat it as unknown rather than as safe.
  • Rate limits are account-level: 60 requests per minute, 1,000 per hour and 10,000 per day by default, surfaced as HTTP 429 with Retry-After and X-RateLimit-* headers.

Frequently asked questions

Can the OpenRush connector change my website, my rankings or my Google data?

No. All eighteen tools are read-only lookups that return data and nothing else. None creates, updates or deletes anything, in OpenRush or in any connected Google property. OpenRush states its analytics reader 'never writes to your property'. The account operations that do mutate state — creating and revoking API keys — exist only as REST endpoints, not as MCP tools.

How much does each OpenRush tool call cost?

Between 1 and 40 credits, at $10 per 1,000 credits, so one cent per credit. OpenRush publishes the table: a dataset export is 1 credit, a live SERP lookup 2, a site audit 9, a keyword gap 12, AI visibility 20, an AI citation map 22, and a backlink gap 40 — about forty cents. Credits are deducted when a call starts, not when it succeeds.

Can an AI agent check my OpenRush credit balance before spending?

No, and this is the connector's sharpest edge. Not one of the eighteen tools reads a balance, a quota or a usage record. OpenRush exposes the balance at the REST path /v1/me/credits and in its dashboard billing page, neither of which the MCP surface reaches. An agent can therefore spend down a balance with no tool that could check the cost first.

How many tools does the OpenRush MCP server have?

Eighteen, not the sixteen Anthropic's directory lists. OpenRush's own unauthenticated capabilities endpoint returned eighteen tools on 2026-08-24, and its documentation and published skill file name the same eighteen. The two the directory omits are inspect_ai_visibility and discover_ai_citations, which at 20 and 22 credits are among the most expensive calls in the surface.

What OAuth permissions does the OpenRush MCP server request?

None that narrow anything. The server's RFC 9728 descriptor declares scopes_supported as an empty array, and its authorization server — a hosted Supabase auth project — advertises only openid, profile, email, phone and offline_access. Every one is an identity claim. Because the surface is entirely read-only, no boundary is missing between a routine call and an irreversible one.

What data does OpenRush read from my Google account?

Aggregate Search Console and Google Analytics reporting, once you connect a property in the OpenRush dashboard. That means clicks, impressions, CTR and position from Search Console, and sessions, users, engagement and traffic sources from GA4. OpenRush states it reads aggregate data only and 'never reads user level data or personal information'. Connecting a property is dashboard-only, because Google OAuth needs human consent.

Why did my OpenRush agent spend more credits than I expected?

Most likely because it followed the next_actions chain. Every OpenRush response carries agent-readable follow-up calls, and a 9-credit domain lookup can return a pre-seeded 40-credit backlink gap call. OpenRush's own skill file warns agents to treat next_actions as suggestions 'not work that has already run'. Scheduled weekly or monthly jobs draw on the same single account balance.

Does the OpenRush site audit crawl my whole website?

No — it samples up to 20 pages. The audit_site tool takes a max_pages parameter that defaults to 10 and is capped at 20, so it is a sampled health check rather than a full crawl. OpenRush also states the audit reads raw HTML only: it does not execute JavaScript and does not measure Core Web Vitals, so client-rendered content is invisible to it.

Sources

More Productivity connectors

  • 6sense

    Ask an AI agent about 6sense account insights, buying signals and keyword trends.

  • Adobe Workfront

    Find, create, update and delete Workfront work items from an AI agent.

Browse all MCP connectorsClaude and ChatGPT setup guideBrowse AI skills

Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.

Use it in an agent

Put OpenRush to work.

Connect OpenRush once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio
Call (650) 285-1019Our AI receptionist answers.