Hand the fix to your AI agent.

New · on Pro and Agency

EkhaneMark has an MCP server. Your coding agent reads the feedback your clients pinned, sees what they saw, fixes it in your code, and sends it back to them for review.

claude · ponno-web

>

● ekhanemark · list_feedback(project: Ponno, status: open)

#12 “Can the trial button be bigger? It's hard to see on my phone.”
Hero › Button · iPhone · Safari · 390×844

● ekhanemark · get_feedback(#12)

screenshot (image), selector a.trial-cta, page /

● ekhanemark · start_work(#12)

● Edit src/components/Hero.tsx

- className="h-9 px-4 text-sm"

+ className="h-11 px-5 text-base"

● ekhanemark · mark_ready_for_review(#12, site: Staging, “The trial button is bigger now”)

✓ Rahim gets an email to check it on the site. It closes when he presses Looks good.

From the client's pin to their approval, without copy and paste.

  1. 1

    You ask your agent

    In Claude Code, Cursor or another MCP client: “fix the open feedback on ponno.app”.

  2. 2

    It reads the pins

    Through EkhaneMark it gets each item’s comment, thread, page, element, device and files, and sees the client’s screenshot.

  3. 3

    It changes your code

    In your repository, on your machine. You read the diff and deploy it yourself.

  4. 4

    The client checks it

    Once you’ve deployed, the agent marks it Ready for review with a summary. The client presses Looks good, or Still broken.

Seven tools, on purpose.

Enough to read a pin and hand it back. Nothing that deletes, closes or deploys.

list_projects
Your sites and their snippet keys, so the agent knows which project your code is.
list_feedback
Open items for a project, filtered by status, page or “assigned to me”.
get_feedback
One item: comment and thread, page and element, browser and screen size, files, and the screenshot as an image.
start_work
Moves it to In progress, so the client sees the team is on it.
ask_client
Asks the client when it’s unclear what they want, instead of guessing.
add_comment
Replies in the thread, or leaves a note only your team sees.
mark_ready_for_review
Hands it back with a summary and the site where the fix is, for the client to approve.
request_feedback
Asks your client for feedback on a page, by a day. What they pin comes back as items to fix.
list_requests
Where each request stands: accepted, declined, answered, and how much of it is done.
set_due_date
Gives an item a deadline, yours or the client’s.

Safe by design.

An agent works for you, not for whoever wrote the comment.

  • Your code, never your live siteThe agent edits the repository on your machine. Nothing deploys until you do.
  • Client comments are data, not ordersThe server hands the agent what the client wrote as the client’s words, never as instructions to follow.
  • Tokens you controlEach token belongs to one person and the projects you pick, and you can revoke it from AI agents in your dashboard.
  • Under your nameWhat the agent does appears as yours, Tanvir’s, like your own work. Revoke its token and it stops at once.
  • The client still closes itAn agent can only mark an item ready. It is done when the client says it is.

One command to connect it.

A token from AI agents in your dashboard and the server's address. Nothing to install. Works with Claude Code, Cursor, VS Code, Windsurf and Antigravity.

$ claude mcp add --transport http --scope user ekhanemark https://ekhanemark.com/api/mcp \
  --header "Authorization: Bearer <your token>"

The dashboard gives you this line with your token in it, plus the setup for Cursor, VS Code, Windsurf and Antigravity.

Then ask: “Fix the open EkhaneMark feedback on this site.” In Claude Code there's also a ready-made command, /mcp__ekhanemark__fix_open_feedback.