agentlinkopsAgentLinkOps / Compare
Read-only SEO MCP servers beside AgentLinkOps.
Your agent can already reach Ahrefs, SE Ranking and ContextBolt over the Model Context Protocol (MCP), the open standard that lets an agent call a vendor’s tools from inside a chat. This page puts AgentLinkOps beside them on the questions a buyer asks: what can the agent ask, what happens when I close the chat, what comes back with each answer, and what does it cost as published. Checked September 15, 2026.
Who this is for
You run link building with an agent, or you are about to. Someone on your side asked which MCP server to connect, and the honest answer is that these four do different jobs. Three are data suppliers: the agent asks a question about a domain or a keyword and gets rows from an index. AgentLinkOps monitors the placements you already have and keeps records about them, whether or not anyone is in the chat. Two of the three suppliers feed it.
If the question you want answered is “which tool has the biggest index”, this page will not help. We do not have that measurement and we will not copy a vendor’s.
The four, in one sentence each
- Ahrefs SEO MCP. Ahrefs’ page describes it as a way to “get live Ahrefs data inside ChatGPT, Claude, and other AI tools” for “keyword research, competitive analysis, backlink audits”, and states it “is available for the following Ahrefs paid plans: Lite, Standard, Advanced, Enterprise.” (fetched September 15, 2026).
- SE Ranking MCP. SE Ranking’s documentation calls it “a standardized interface that allows any compatible AI model or agent to access your SE Ranking data”, and says the API key behind it “powers Data/Research tools (keywords, backlinks, domain overview, SERP, AI Search) and Project Management tools (rank tracker, audits, competitors, prompt monitoring)” (fetched September 15, 2026).
- ContextBolt SEO. ContextBolt’s page offers “keywords, SERPs, backlinks and AI visibility, all from one chat” to any client that speaks MCP, and publishes a per-tool credit table beside a flat monthly price (fetched September 15, 2026).
- AgentLinkOps. A hosted service and a local CLI that verify the source and destination URLs you supply, record dated observations with evidence, keep checking on a schedule with nobody connected, and deliver changes as an event feed and signed webhooks. Hosted access is invitation-only.
Side by side
Competitor cells quote the vendor’s own page as fetched on September 15, 2026. AgentLinkOps cells quote the public claims ledger; the ledger row names the source file behind each one. “No statement found” means we did not find one on that vendor’s fetched page that day, and nothing more.
| Question | AgentLinkOps | Ahrefs SEO MCP | SE Ranking MCP | ContextBolt SEO |
|---|---|---|---|---|
| What the agent can ask | Verify a supplied source page against a target, read a placement’s history and evidence, import a backlink export you already own, extract published contact routes from a saved page, and read change events. It does not hold a web-wide index: reports describe your monitored dataset only, and no gap report says a competitor has a link you lack. | “Keyword research, competitive analysis, backlink audits” over live Ahrefs data, including “backlink profiles” and “link-building velocity” (their page, fetched September 15, 2026). | Data and research tools for “keywords, backlinks, domain overview, SERP, AI Search” plus project tools for “rank tracker, audits, competitors, prompt monitoring” (their documentation, fetched September 15, 2026). | “Keyword research, keyword difficulty, Google SERP results, domain and competitor analysis, backlink tools” and whether AI answers name you (their page, fetched September 15, 2026). |
| Do checks keep running after the chat closes? | Yes. Every hosted watch carries a cadence and a next-check time, and the service’s own five-minute schedule dispatches whatever is due with nobody connected. Changes land in a resumable event feed and in signed webhooks on their own ten-minute schedule. | A third-party review dated July 16, 2026 states: “It only works inside a chat you start. Close the window and nothing happens.” Ahrefs’ own MCP page makes no statement about scheduling or alerts (fetched September 15, 2026). | No statement found on the MCP documentation page about the server itself scheduling checks or alerting. SE Ranking’s rank tracker and audits are its own platform features and are outside this comparison. | Their page says “every lookup your agent runs lands here on its own, dated and sorted” and “your research saves automatically to your dashboard”. No statement found about scheduled checks or alerts (fetched September 15, 2026). |
| Evidence with each answer | Each observation carries the fetched URL and every redirect hop, the HTTP status, the robots posture, a SHA-256 hash of the bytes read, and a standing warning that JavaScript execution and visual visibility were not checked. Hosted rows keep the hash beside the result; the retained snapshot is readable until retention removes it, at which point the call says EVIDENCE_EXPIRED. | Index rows, capped per request by plan: 100 on Lite, 250 on Standard, 500 on Advanced, unlimited on Enterprise (their MCP page, fetched September 15, 2026). No statement found about fetch evidence behind a row. | Index rows metered per record. No statement found about fetch evidence behind a row. | Lookups are saved “dated and sorted” in the dashboard (their page, fetched September 15, 2026). No statement found about fetch evidence behind a row. |
| Can the agent read the meter? | Yes. get_usage reports reserved, consumed and released as three separate totals, and a unit is consumed only when its observation is saved. The same call names the enforced monthly ceiling and what remains of it. | The MCP page lists API units per month by plan: 200,000 on Lite, 800,000 on Standard, 2M on Advanced, custom on Enterprise. The July 16, 2026 review states each call draws from that allowance (fetched September 15, 2026). No statement found about a call that reads the remaining balance. | Their credit page states “every successful call (2xx status) consumes a number of credits from your balance” and “failed requests (4xx or 5xx) do not consume credits”; a backlink summary costs 100 credits per record and a backlink export 1 credit per record; the subscription endpoint that reads the balance costs 0 credits. The page does not say whether the MCP draws the same credits (fetched September 15, 2026). | Their page states “different tools cost different amounts, and one question can trigger several calls”, then prices them: keyword, SERP, domain and competitor lookups 1 credit; backlink overview, referring domains and gap 3 credits (fetched September 15, 2026). No statement found about a call that reads the remaining balance. |
| Billing model as published | No public price. Hosted access is an invitation-only pilot, and payment collection does not exist in this build: there is no checkout, and nothing on this site can be charged. Usage is metered in check units you can read before any bill exists. | Seat plus units. Lite $129, Standard $249, Advanced $449 and Enterprise $1,499 a month with an annual commitment; “1 user included” on each, extra users $40, $60 or $80 a month by tier; every plan lists “API and MCP access” (their pricing page, fetched September 15, 2026). | Subscription plus credits. Core $129 and Growth $279 a month billed monthly, with 25K and 100K API credits a month respectively and “API credits & MCP access” on both; “extra manager seats from $16” (their pricing page, fetched September 15, 2026). | Flat monthly credits. “$35 a month with 1,000 credits, after a 7-day free trial that includes 100 credits”; card required; month to month; top-up packs from $9 (their pages, fetched September 15, 2026). |
| What it is for | Watching placements you already earned, with a dated record you can inspect and export, and a schedule that outlives the session. | Asking Ahrefs’ index questions from inside a chat, on a paid Ahrefs plan (their page, fetched September 15, 2026). | Asking SE Ranking’s data and project tools questions from inside a chat, on an SE Ranking plan (their pages, fetched September 15, 2026). | Index lookups at a flat monthly price with a published per-tool credit table (their pages, fetched September 15, 2026). |
Ahrefs and SE Ranking as suppliers
AgentLinkOps reads the exports these servers’ parent products produce. An Ahrefs backlink export imports through a named preset. An SE Ranking export imports through the generic CSV path with an explicit column map, because there is no SE Ranking preset today; the map is the mechanism and a preset is only a convenience over it. Either way, an import preview writes nothing, lists every rejected row with its line and reason, and never claims completeness.
That is why this page calls them suppliers. The rows they return are where a campaign starts; the placements you keep are what AgentLinkOps watches. The Ahrefs import reference and the generic CSV reference show the columns each path expects.
The one claim we make
An MCP directory’s roundup of SEO servers, marked updated June 2026, says of the category: “these servers are almost all read-only. They analyze; they don’t publish, schedule, or alert.” That is their summary on that date, and it matches what we found on the three fetched pages: no statement that a check runs after the chat closes.
AgentLinkOps makes one comparative claim, and it is checkable in the source: scheduled checks and change events continue after the chat closes. A hosted watch carries a cadence and a next_check_at; the service’s five-minute schedule dispatches due watches on its own; list_link_events returns a page and an opaque cursor; signed webhooks deliver on their own ten-minute schedule. That claim is about scheduling existing in the deployed code shape. It is not an availability figure; no percentage appears here until a status page measures one.
If you have an invited account, ask your agent to prove it with four read-only calls.
What this page does not claim
- No tool counts. The vendors publish different counts for the same server on different pages, and a count says nothing about what a call returns. Our own tool catalog renders from the registry at build time for the same reason.
- No index size, freshness or accuracy figure for any vendor. We did not measure one.
- No statement that any of the three cannot do something. “No statement found” is the strongest thing this page says about an absence.
- No live discovery on our side. Requests for new supplier discovery stop before persistence in this build with
PROVIDER_NOT_CONFIGURED. Imports of exports you already own work today. - No open signup and no price. Reading this page or the public brief creates no account and starts no monitoring.
How this was checked
Written September 15, 2026 against the repository at commit fa927cf. Each AgentLinkOps statement quotes a row of the DP-0023 claims ledger (W-02, W-03, W-07, W-08, M-02, M-04, M-06, A-01, A-03, A-10, P-01 to P-04 and P-06), and each row names the source file and line or the dated record behind it.
Competitor statements were fetched on September 15, 2026 from these pages and are recorded as ledger rows C-01, C-02 and C-04 to C-10: the Ahrefs SEO MCP page and Ahrefs pricing page; the SE Ranking MCP documentation, pricing page and API credit system page; the ContextBolt home page and ContextBolt SEO page; the MCP.Directory roundup marked updated June 2026; and a third-party Ahrefs MCP review dated July 16, 2026. SE Ranking’s figures are the US-dollar prices its pricing page showed that day.
A competitor’s number is their published statement on the fetch date, never a capability we verified. If a vendor page changes, the cell that quotes it is stale until this page is rechecked and its date updated.