Skip to content
Why GEO Why us Pricing
Get started free
HomeProductIntegrations & API

A number nobody sees changes nothing.

Integrations are the ways a number leaves the console: REST API, MCP server, Slack and email digests, CSV and PDF exports, CMS publishing, and SSO.

Read the API docs Get started free →
The Reading the data group, 3 connected: Search Console read 23 Jun 2026, Google Analytics 4 read 23 Jun 2026, and crawler hits receiving logs.
What each connection last read, and when it last read it.

Three ways in

Integrations and API is how Praised data leaves Praised: a REST API with scoped keys, an MCP server for assistant clients, Slack and email digests after every scheduled run, CSV and PDF exports, direct CMS publishing, and OIDC single sign-on you configure yourself. A visibility number only changes behaviour if it reaches the meeting where decisions get made, and that meeting is rarely held inside our console. Pull it, get it pushed to you, or ask an assistant about it. Same data, same permissions.

The integrations list in focus: reading the data (Search Console, Google Analytics 4, crawler hits), reading the workspace (REST API keys, MCP server, Looker Studio) and publishing (WordPress, Slack webhook, report digest), each with its state.
Demo workspaceNine connections in three groups, each showing what it last read and what it feeds. One that stops reading turns its figures into “couldn’t check”, never into zero.

REST API

Runs, per-prompt results, citations, competitors, experiments and the action queue — all readable with a workspace-scoped key you create and revoke yourself.

Enough to build your own dashboard, feed a warehouse, or wire visibility into an internal scorecard.

On every paid plan, from $99. No enterprise gate, no access request: name a key in Settings → API keys and it is live.

MCP server

Point Claude, Claude Code or any MCP client at the server and ask your own data questions in plain language — which prompts slipped, what is open in the queue, what changed since last run.

Twenty-two tools: eighteen read — the measurements, the approved facts, the format contracts — and four work the queue or hand a draft back. All under your own token, scoped to what you can already do.

Push notifications

A summary posted to your Slack channel or webhook after every scheduled run, drift alerts when a tracked rate moves beyond its interval or a rival overtakes you, and an email digest on the cadence you choose.

Slack delivery is a webhook URL you paste once — no app to install, nothing to grant.

Everything that connects

All of it ships on every paid plan, and none of it needs asking for. There is no integrations tier: what separates the plans is credits and how many products you track, not which doors we unlock.

Engines measured

Six answer surfaces, none plan-locked — credits are the only limit.

ChatGPTClaudeGeminiPerplexityGoogle AI OverviewsGrok

Works inside your AI tools

One MCP server, 22 tools, any MCP client — write where you already work, ground and grade it here.

ClaudeClaude CodeChatGPTCursorAny MCP client

Your data in

Search demand and AI-referred sessions synced daily; AI-crawler hits shipped from your edge with a copy-paste snippet.

Google Search ConsoleGoogle Analytics 4CloudflareVercelNetlifyAny server log

Numbers out

A summary on your channel after every run; the digest to whoever the report is for — including people without a login.

SlackWebhooksEmail digestExternal recipients

Publishing

Three CMS connectors that write drafts and only drafts — plus clean exports for everything else.

WordPressSanityContentfulHTMLMarkdownJSON-LDllms.txt

Reporting & BI

The REST API and CSV feed any BI tool today; the Looker Studio connector is in final checks before its gallery listing.

REST APICSV exportPrintable reportLooker Studio

Sign-in

Self-serve OIDC and SAML — issuer, certificate, domain allowlist. No sales call to turn either on.

OktaMicrosoft Entra IDGoogle WorkspaceAny OIDC providerAny SAML IdPSCIM provisioning

Every item, every paid plan — see pricing for what does differ (credits and tracked products).

For agencies and multi-brand teams

A workspace per client, credits pooled across all of them, and a portfolio view that ranks every brand you manage on one screen.

  • Switch workspace without signing out
  • Role-based access, per workspace
  • Reports you can hand to a client under your own brand
  • One API key scope per workspace, revocable independently

One subscription covers the roster rather than one per client, so the credits follow the work instead of the invoice. How agencies run it →

What we ask of your stack

Nothing, to start. No tag, no script, no DNS record, no CMS connection. Measurement reads the public web the way an engine does, so a first run needs only a domain.

Connections exist to remove steps later — publishing without a copy-paste, alerting without a check-in — never as a prerequisite for the number.

API and MCP reference →

Setting up the two you install

The API is a key away and the digests are a checkbox. These two live outside the console, so here is the whole setup.

MCP server — your data inside an assistant

Point Claude Code, Claude Desktop, Cursor or any MCP client at it and ask in plain language. Twelve tools read the measurements; three work the fix queue.

claude mcp add praised \
  --env PRAISED_API_KEY=<your key> \
  -- npx -y @praised/mcp-server

The key comes from Settings → API keys and does not expire; revoking it cuts the integration off at once. It is traded for a short-lived token behind the scenes, so nothing permanent travels to the measurement API.

  • list_projects
  • get_project
  • list_runs
  • get_run_metrics
  • get_trends
  • get_citations
  • get_benchmark
  • get_hallucinations
  • get_experiments
  • list_actions
  • get_fix
  • check_ai_access
  • run_site_audit
  • mark_action_fixed
  • verify_actions

Everything runs under your own token, scoped to what you can already see and do. Marking an action fixed triggers an immediate re-check, and a finding that is still there goes back on the queue — an assistant cannot mark its own homework.

Looker Studio — the trend in the monthly report

One row per metric per run, with the interval bands intact, in the report you already send. The connector is built and in final checks before its gallery listing; until that lands, the REST API and CSV export feed Looker Studio through its native connectors.

Run date · Metric · Run ID · Change beyond noise · Value · CI low / CI high (95%) · Sample size (n) · Delta vs previous run

  1. Once the gallery listing is live: open the connector, authorise it, and paste an API key from Settings → API keys.
  2. Today: export any view as CSV, or pull /v1 endpoints into Looker Studio's Sheets or CSV connectors with a key from the same place.
  3. Chart Value by Run date, broken down by Metric, with the CI fields as bands.

A missing interval stays empty rather than becoming a zero — a chart should never draw certainty we did not measure.

Questions about connecting

What can I do with the MCP server?
Point an MCP client such as Claude or Claude Code at it and ask your measurement data questions in plain language — which prompts you lost this week, what changed since the last run, what is open in the fix queue. It can also work the queue: run a site audit, mark an action fixed, re-verify. Marking something fixed triggers an immediate re-check, and a finding that is still there goes back on the queue — an assistant cannot mark its own homework. Everything runs under your own token, scoped to what you can already see and do. Setup is here.
Is the API available on every plan?
Yes — everything on this page ships on every paid plan, from Pro at $99. There is no integrations tier: plans differ in credits and how many products you track, not in which doors we unlock. It is also self-serve end to end — you name a key in Settings → API keys, it is shown once, and the same key drives the REST API and the MCP server. Nobody has to approve it and there is no form to fill in. Category norms as of August 2026 put the API on the enterprise plan and behind an access request, so it is worth asking whoever else you are evaluating what their API actually costs to switch on.
Which CMS can you publish to?
WordPress, Sanity and Contentful directly, and everything else by export. WordPress needs no plugin — an application password and the built-in REST API are enough. Sanity and Contentful take a structured document against a content type you define, so you map our fields onto your field names; a required field nothing maps to is named in the preview as skipped rather than guessed at, and a field you did not map is not written at all. Everything else goes out as clean HTML, Markdown or JSON-LD, with the JSON-LD inline. Every destination receives DRAFTS — drafts in Sanity, an entry never published in Contentful, a draft post in WordPress — because publishing to a live site is yours to do after you have read what we wrote.
Do you support single sign-on?
Yes — self-serve OIDC and SAML, and SCIM provisioning to go with them. You configure your issuer or IdP metadata and certificate in workspace settings yourself; none of it requires a sales conversation or an onboarding ticket.
Can Praised push data into a system you have no connector for?
Yes, through a flow. A Call step sends a request to any endpoint you name — your own service, a Zapier or Make hook, an internal queue — as part of a graph that can run on a schedule. The address is fixed by whoever builds the flow and the running model can never read or change it, which is the guard that matters when an agent reads competitor pages and open-web content by design. An injection can influence what a call sends; it can never influence where.

Keep going

API & MCP docs
Endpoints, auth and install, before you sign up
Pricing
Which plan includes what
Flows
Call an endpoint we have no connector for, on a schedule
Platform overview
All ten surfaces in one place

Wire it into where you already work.

Start free, connect what you need later. The first run asks for a domain and nothing else.

Get started free Read the docs