dot.

Turn a Claude artifact or preview URL into a shareable review link and read pinned feedback.

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 dot. connector does

The dot. MCP server turns something you built — a Claude artifact or a deployed preview URL — into a shareable review link, then reads the pinned comments back into the conversation. Three tools: one creates projects and links, two read projects and feedback. Sign-in is OAuth 2.1 with PKCE through Google or GitHub.

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.

dot. tools (3)

  • create_review_link
  • list_projects
  • get_feedback

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

Limits

  • Review links are public by default and free accounts cannot restrict them. dot.'s own application code disables the restricted_review visibility option unless the account is on a team or a paid plan. Anything an agent publishes on a free account is readable by anyone with the URL.
  • The connector cannot set visibility. dot. documents no visibility parameter on create_review_link, and sharing is managed in the dashboard. Even on a paid plan, restricting a link is a separate manual step after the agent has created it.
  • Scopes do not separate reading from writing in practice. read and write are advertised together and issued together, so consenting to read feedback also consents to creating projects and public links. No boundary falls between the routine read and the publishing write.
  • We could not read tool schemas or safety annotations. The endpoint returned 401 to an anonymous request, so no readOnlyHint or destructiveHint is reported here, and parameter names beyond "a URL or code" are not published.
  • The free-plan quota is published two different ways. Terms say three active review links and unlimited projects; the pricing page says one shared prototype and feedback on three screens. We could not reconcile them from public sources.
  • No tool can observe quota or billing. dot. states Claude cannot access billing information, and no tool name carries a credit, usage or quota verb. An agent cannot check its remaining allowance before spending it.
  • The URL proxy does not render every page. dot.'s terms state it "is designed for reviewing static and server-rendered pages" and that "JavaScript-heavy single-page applications may not render correctly." Localhost URLs cannot be reached at all.
  • The owner PIN is stored in plaintext. dot.'s privacy policy discloses this for the optional 4-digit owner-view code and calls it "a convenience feature, not a security credential."
  • dot.'s documentation is unreadable to most text extractors. app.leaveadot.com is a Vercel single-page app that returns the same 713-byte HTML shell for every path, including robots.txt, llms.txt and a nonexistent control path. The documentation text exists only inside the JavaScript bundle. Any tool fetching the doc URL as HTML sees an empty page.
  • Access tokens expire after 90 days. dot.'s privacy policy states this for MCP integrations. A long-lived automation will stop working and needs reconnecting.

Frequently asked questions

Who can open a dot. review link the connector creates?

Anyone who has the link, with no sign-in, unless you are on a paid plan and change it. dot.'s dashboard offers two visibility settings, public and Only invited people, and its own code disables the restricted option for free accounts without a team. The tool description mentions only the public outcome.

How many tools does the dot. MCP server have?

Three, and dot. publishes the same three. Anthropic's directory lists create_review_link, list_projects and get_feedback; dot.'s Claude documentation lists identical names with read or write labels, and its Codex guide repeats them. Both enumerations match as literal strings in both directions, so nothing is hidden behind a fourth name.

Can the dot. connector delete a project or edit feedback?

No, and dot. states the boundary directly. Its Claude documentation says Claude cannot delete projects, modify feedback, or access billing information. The connector grants create on projects and review links, plus read on your projects list and feedback comments. Deletion stays in the dashboard, where deleted projects are purged after 30 days.

Does creating a dot. review link send anyone an email?

Not the link itself, but the project it creates becomes an email source. dot.'s terms list email notifications for new feedback and reply alerts as part of the service, delivered through Resend. Once a reviewer comments, owner digest emails follow, and reviewers who opt in receive reply notifications.

What OAuth scopes does the dot. MCP server request?

Two, read and write. dot.'s RFC 9728 resource descriptor and its authorization server metadata both advertise exactly those, checked live on 2026-08-23. They name no object, so the grant cannot separate reading feedback from creating projects. Tokens expire after 90 days per dot.'s privacy policy.

Does the dot. connector cost money to use?

No call charges a fee, but the free plan caps review links and no tool reports the cap. dot.'s terms give free accounts three active review links and 30-day feedback history; its pricing page says one shared prototype and feedback on three screens. The two published limits disagree.

Can Claude read feedback written by people outside my team?

Yes, and that is the connector's purpose. get_feedback returns comment text, author name, timestamp and thread structure. Reviewers need no account, and dot.'s privacy policy notes the self-provided name is not verified and may be a pseudonym. Treat returned comment text as untrusted input rather than instructions.

Why does a localhost preview fail with the dot. connector?

Because dot. loads the page from its own servers, not your browser. Its Codex documentation states dot. cannot reach localhost and tells you to deploy a preview to Vercel, Netlify or similar first. The proxy also targets static and server-rendered pages, so JavaScript-heavy single-page apps may not render correctly.

Sources

More Developer Tools connectors

  • Base44

    Build and edit Base44 apps from an AI assistant, and query data your app's users submit.

  • CircleCI

    Diagnose failed builds, read job logs, inspect test results and rerun or roll back deployments from your AI assistant.

  • Claude Managed Agents

    Create, run, monitor and debug Anthropic's Claude Managed Agents from inside Claude itself.

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 dot. to work.

Connect dot. 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.