MCP and API

Your anewera data in every AI tool

With the anewera MCP server you ask Claude, ChatGPT or your own agent about your AI visibility directly: shares with their margin of error, answers, sources, competitors and actions. And if you like, your agent keeps questions, competitors and projects up to date, as you do in the dashboard.

At a glance

MCP server address

https://app.anewera.ai/mcp

The previous address https://app.anewera.ai/api/mcp keeps working.

  • Included

    In the standard package at CHF 290 a month and in the 7-day trial, at no extra cost

  • 120 requests a minute

    Per key or connection on the package, 60 in the trial

  • Data in Zurich

    Stored in Switzerland. A connection sees only what its person sees in the dashboard

What you can ask

You ask in your own words. Your AI tool calls the right anewera tools and answers with the numbers from your dashboard.

  • How visible are we in ChatGPT this week, with the margin?

  • Which sites cite our competitors but not us?

  • Show me Claude’s answers to our most important question.

  • Which searches does Gemini run before it answers?

  • Write a brief for the top content action.

Address and sign-in

MCP (Model Context Protocol) is an open standard through which AI tools read data from other services. Every tool connects to the same address, in one of two ways.

Sign in (OAuth 2.1)

Your tool opens the anewera sign-in. There you choose the organisation and the access: “Read only” or “Read and change”. The tool never sees your password.

For Claude, ChatGPT, Cursor, Gemini CLI, VS Code and most other programs.

Personal API key

For tools and scripts that send a header. You make the key (anw_...) in the app on the Connections page. It travels in this header:

Header
Authorization: Bearer anw_...

For Claude Code, Codex CLI, your own scripts and the REST API.

Setup for each tool

A few steps, with a sign-in or a key depending on the tool. Where it says anw_..., put in your own key.

Claude

claude.ai and Claude Desktop · With sign-in

  1. In Claude, open Settings, then Connectors, then Add custom connector.

  2. Name: anewera. URL: https://app.anewera.ai/mcp

  3. Click Connect, sign in to anewera and choose the organisation and the access.

Claude Desktop and the mobile apps share the same connector. On Team and Enterprise plans an owner adds it for everyone, then each person connects.

Claude Code

In the terminal · With sign-in or API key

  1. Run the command in the terminal:

    Terminal
    claude mcp add --transport http --scope user anewera https://app.anewera.ai/mcp
  2. In Claude Code, type /mcp, choose anewera and Authenticate. Your browser opens the anewera sign-in.

With an API key instead of the sign-in

With an API key instead of the sign-in, add the header to the command:

Terminal
claude mcp add --transport http --scope user anewera https://app.anewera.ai/mcp \
  --header "Authorization: Bearer anw_..."

ChatGPT

In developer mode · With sign-in

  1. In ChatGPT, open Settings, then Apps and Connectors, and switch on developer mode under Advanced settings.

  2. Click Create. Name: anewera. MCP server URL: https://app.anewera.ai/mcp. Authentication: OAuth.

  3. Sign in to anewera and choose the organisation and the access.

On Business and Enterprise plans an admin sets it up for the workspace.

Codex CLI

In the terminal · With API key

  1. Make an API key on the Connections page.

  2. Add anewera to the file ~/.codex/config.toml:

    ~/.codex/config.toml
    [mcp_servers.anewera]
    url = "https://app.anewera.ai/mcp"
    bearer_token_env_var = "ANEWERA_API_KEY"
  3. Set the key in your shell:

    Terminal
    export ANEWERA_API_KEY="anw_..."

Codex reads the key from the environment variable ANEWERA_API_KEY.

Cursor

In the editor · With sign-in or API key

  1. Add anewera to the file ~/.cursor/mcp.json:

    ~/.cursor/mcp.json
    {
      "mcpServers": {
        "anewera": {
          "url": "https://app.anewera.ai/mcp"
        }
      }
    }
  2. Cursor shows Needs login next to anewera. Click it and sign in to anewera.

With an API key instead of the sign-in

With an API key, add the header. Cursor reads the key from the environment variable ANEWERA_API_KEY:

~/.cursor/mcp.json
{
  "mcpServers": {
    "anewera": {
      "url": "https://app.anewera.ai/mcp",
      "headers": {
        "Authorization": "Bearer ${env:ANEWERA_API_KEY}"
      }
    }
  }
}

Gemini CLI

In the terminal · With sign-in

  1. Run the command in the terminal:

    Terminal
    gemini mcp add --transport http --scope user anewera https://app.anewera.ai/mcp
  2. Start Gemini CLI and type:

    Gemini CLI
    /mcp auth anewera

Your browser opens the anewera sign-in.

VS Code

In agent mode in chat · With sign-in

  1. Create the file .vscode/mcp.json in your project:

    .vscode/mcp.json
    {
      "servers": {
        "anewera": {
          "type": "http",
          "url": "https://app.anewera.ai/mcp"
        }
      }
    }
  2. Sign in to anewera when VS Code asks. Agent mode in chat then uses the anewera tools.

The server’s tools

The MCP server offers these tools, and your AI tool picks the ones that fit your question. Without a period they read the last 14 days; for another one they take from and to or days, and engines for single AI services.

Read

list_projects
The projects you can see: the brand, the number of questions and when the AI services were last asked. The other tools take a project from here.
get_overview
Your brand’s headline numbers: visibility, share of voice, named first and recommended, each with its margin of error and k of n answers. With the change against the period before and the daily trend.
get_competitors
Every tracked brand, the most visible first, with the same numbers.
get_source_gaps
Sites cited for your competitors, not for you.
search_answers
Full-text search in the AI services’ answers.
get_answer
One answer in detail: the brands it names and how, the sources it cites and read, and the searches the AI service ran for it.
get_questions
A project’s questions, with your brand’s visibility on each AI service.
get_question
One question in detail: per AI service the brands named, the searches, the cited sites and the newest answers.
get_fan_outs
The AI services’ own web searches before they answer.
get_engines
Per AI service: answers, failures, web searches and the models.
get_table
The dashboard’s tables as CSV.
get_action_prompt
For one action: a prompt for a coding agent or a brief for a writer.
get_visibility
One brand, your own unless you name another, on every question and AI service: named, named first and recommended, each with its margin of error and k of n answers.
get_trend
Visibility day by day, pooled and per AI service, with what moves the numbers: questions added or archived, actions done and model changes.
get_share_of_voice
Each brand’s share of voice, pooled, per AI service and per day.
get_perception
What the AI services say about your brand: one week’s fact check, objections and qualities.
get_location
The location test: the same questions asked from other countries.
get_project_settings
A project’s settings: countries, languages, schedule, time zone, AI services and your facts.
get_plan
Your plan and what of it is used: questions, projects, competitors, countries and requests a minute.
get_report_link
Links to the report as a PDF, for the period, this month and last month.
get_question_settings
A project’s questions as on the settings page, active and archived, with what your plan has left.
get_brands
Your brand and the competitors with their spellings and websites, the archived ones apart.
get_share_links
A project’s share links, for owners and admins.
get_drafts
The draft emails among the actions. Nothing is sent.

Change, with “Read and change”

set_action_status
Sets an action’s status, for example to done.
add_questions
Adds one or up to 100 questions at once, all of them or none, within your plan.
edit_question
Changes a question. A new wording becomes a new question, and the old one is archived with its answers.
archive_question
Archives a question. Its answers stay.
restore_question
Brings an archived question back with its answers.
add_competitor
Adds a competitor, with its spellings and websites.
add_own_brand
Sets the own brand when a project has none yet.
edit_brand
Changes the name, spellings or websites of your brand or a competitor.
archive_competitor
Archives a competitor. Earlier mentions stay.
restore_competitor
Brings an archived competitor back.
update_project
Changes a project’s name, countries or languages, for owners and admins.
create_project
Makes a new project within your plan, for owners and admins.
create_share_link
Makes a share link to read without signing in, for owners and admins.
revoke_share_link
Revokes a share link, at once and for good.
discard_draft
Discards a draft email among the actions, for owners and admins.

Changing over MCP and the API

With the access “Read and change”, your AI tool changes the same as you do on the settings pages and among the actions, if your role is owner, admin or member. The REST API makes the same changes with POST, PATCH and DELETE.

  • Add questions, one or up to 100 at once, and change, archive and restore them
  • Add, change, archive and restore the own brand and competitors
  • Change a project’s name, countries and languages, and make a new project, as an owner or admin
  • Make and revoke share links, as an owner or admin
  • Set an action’s status and discard a draft email

anewera refuses anything beyond your plan, with the same sentence as the dashboard. Nothing is deleted: questions and competitors are archived and keep their answers. Every change is recorded with the person, the key or connection, the tool and the time.

For example:

  • Add these five questions to the project.

  • Archive the competitor “Example Ltd”.

What stays in the dashboard

You change these only yourself in the dashboard, never over MCP or the API:

  • Anything that costs extra or changes the subscription: billing, the package and an extra measurement
  • People and their roles
  • Making or revoking keys and connections
  • Your account, the password and deleting an organisation

REST API

The same data as JSON, for your own scripts, sheets and reports, with the same keys and OAuth tokens as the MCP server.

Fetch your projects. The example reads the key from the environment variable ANEWERA_API_KEY:

Terminal
curl -H "Authorization: Bearer $ANEWERA_API_KEY" \
  https://app.anewera.ai/api/v1/projects

Security

What a connection sees and may do, how long it lasts and how you end it.

  • Sees what you see

    A connection or key sees exactly what its person sees in the dashboard, in one organisation only.

  • Reads unless you allow more

    A connection can change something only if you choose “Read and change” and are an owner, admin or member. Every change is recorded.

  • Unknown applications

    An application anewera does not know, because it returns to an unknown address, shows as unverified with its address and gets read access only.

  • Short lives

    Access tokens last one hour. Refresh tokens are replaced at every use; if one is used twice, the connection ends.

  • Stored as hashes only

    We store keys and tokens only as SHA-256 hashes.

  • Revoked at once

    On the Connections page you revoke every connection and key, with immediate effect.

  • No password in the tool

    You sign in to anewera, not in the tool. The application never sees your password.

  • Data in Zurich

    Your data stays in Zurich, Switzerland.

  • Answers are third-party content

    AI services and websites write the answer texts. The tools say so, so that agents treat them as data, not as instructions.

  • No paid measurement

    No tool starts a paid measurement.

Manage connections and keys

Let us find out what AI says about you.

In the first meeting we look at your market together. Free, 30 minutes, no obligation.

Book a first meetingor the free AI check