LeaderboardEvery site, rankedFree toolsChecks, no sign-upAgentsYour agent fixes itDocsThe API and MCP

Agentic SEO for every site your team runs

One rulebook across all of them, an API and MCP server your agents already speak, and an expert to plan the work with you.

Book a strategy sessionRead the API
TerminalSample
$ curl -s https://mttr.si/api/v1/sites/acme.example/changes \
    -H "Authorization: Bearer $MTTR_KEY"
{
  "changes": {
    "score":  { "before": 78, "after": 84 },
    "fixed":  [{ "rule": "canonical", "was": 31 }],
    "broke":  [],
    "pages":  { "before": 96, "after": 100 }
  }
}
$ jq -e '.changes.broke | length == 0'   # fail on a broken rule
Step 1 of 8What would you like out of it?

Pick any. The call is built around these.

Step 2 of 8Which site matters most?

We read it before the call. If mttr has read it already, you will see where it stands.

Step 3 of 8How many sites do you run?

Yours, or the ones you run for clients.

Step 4 of 8Whose sites are they?

And the line of work, if one fits.

Step 5 of 8How are they built and run?

So the plan fits how your team works.

Built with
Agents you use
Visits a month
Step 6 of 8When would you like to start?

There is no wrong answer. Exploring is fine.

Step 7 of 8Who are we talking to?

Used only to arrange the call.

What you do
Step 8 of 8Your session, as we will plan it

Check it over, add anything we should know, and send.

What working with us looks like

For every site your team runs, from the first read to the plan, in five steps.

  1. ReadEvery site you run is read against one rulebook, every page, the way search engines and AI assistants read it.
  2. RankEach site gets its fixes ranked by what they are worth. The same template problem across nine sites is fixed once, in the template.
  3. Hand offYour agents take the fixes over MCP or the API, in the tools your team already uses, and you decide what ships.
  4. CheckAfter a release, mttr reads again and says what moved and what broke, so a bad deploy is caught before anyone outside sees it.
  5. PlanOne of our experts goes through it with you and writes the plan: the fixes worth the most, in order, for your team or its agents.
Fixes worth the most, across your sitesSample
  1. Add a tap-to-call link9 sites, 412 pages+6
  2. Answer in the first lines14 sites, 860 pages+4
  3. Describe every page22 sites, 1,310 pages+3

Together, on each site they reach+13

MCP
https://mttr.si/mcp, 62 tools over Streamable HTTP. Claude Code, Codex, Cursor, Gemini CLI and any MCP client connect in one line.
REST
JSON under /api/v1, described in OpenAPI, every call in the reference.
In CI
After a deploy, POST /sites/{site}/rescan, then fail the build when /changes lists a rule that broke.
One rulebook
Every page of every site checked against the same rules in four stages: access, understand, trust and act. How it grades.
Rivals
The sites doing the same work, found by what their pages say and where they are: mttr_competitors, mttr_gaps, mttr_market.
History
Every read kept, and what each changed: rules fixed, rules broken, pages added and gone.
SOC 2
Our report is available on request.
Decisions
Code passes or fails a page. A model only advises, and anything it is unsure of goes to a person.
Privacy
No cookies on public pages. Nothing follows a visitor.