Does Marker.io have an MCP server?

Yes, and it is a good one. It is generally available, included on every plan at no extra cost, and gives an agent the full bug payload including console logs and network requests, so the agent can investigate, comment, attempt a fix and open a pull request. Claude Desktop, Claude Code, Cursor, Windsurf and any MCP-compatible client are supported.

Then why would I use Writhink's MCP instead?

Because the input is different, not because the server is better — having an MCP server is table stakes in this category now. Marker.io's payload describes a technical failure. Writhink's describes what a non-technical client wants changed, with the pin, the element, the breakpoint and the screenshot, and the prompt layer rewrites, merges and prioritizes those comments before any agent sees them. If your feedback is bugs, Marker.io. If it is client opinions, Writhink.

Do clients need an account with Marker.io?

Not for the reporter path — customers and end users submit through the widget and follow their own issues with no account. Guests, the richer collaborator role with portal access, are a counted and capped seat on each tier; the table above has the counts and the date we checked them. Writhink has no equivalent count; anyone with the link comments with a display name.

Can Writhink send feedback to Jira?

No. There are no native integrations of any kind. What you get instead is the generated prompt — copy it or download the .md — and the MCP server. An agent connected over MCP can read the feedback and act on it, but it will not create a Jira issue for you.

Which should I use for a release QA pass?

Marker.io. A QA pass produces reports that need reproduction steps, console output and environment detail, and it should land in your tracker. Writhink is for the client round before or after that, where the input is language rather than stack traces.

Comparison

Writhink vs Marker.io: who is filing the report changes everything

Marker.io turns a bug report into something a developer can act on immediately: annotated screenshot, console logs, network requests, browser, OS, session replay, filed straight into Jira or Linear. Writhink starts from a different person — the client who says the spacing looks off — and ends somewhere different, with one prompt your AI coding assistant can run. Two good tools for jobs that only look the same.

See what your AI gets

No credit card. Your client doesn’t need an account.

What Marker.io is good at

Nothing here matches Marker.io on developer handoff. A report arrives with the annotated screenshot plus console logs, network requests, the URL, the browser and the OS, and session replay with data masking sits behind it. That is the difference between a bug you can reproduce and a bug you have to chase over email.

It is designed to feed the tracker you already use rather than replace it: Jira, GitHub, GitLab, Azure DevOps, Linear, Asana, ClickUp, Trello, Teamwork, Notion, Slack and more. Three installation paths — a code snippet, a WordPress plugin or a browser extension — mean it fits sites you cannot modify as well as ones you can, across all browsers and mobile.

Their MCP server is generally available, included on every plan at no extra cost, and hands an agent the whole payload: screenshot, URL, console logs, network requests, browser and OS, and the user's description. That is the credible version of the agent investigates, fixes and opens a pull request story, and for a technical bug it is richer than what Writhink's server carries. They also run a free trial with no credit card and a discounted agency plan, which says they understand the agency buyer.

What arrives with one report

	Writhink	Marker.io
Screenshot	Automatic, of the pinned spot	Annotated screenshot
Where on the page	Pin position, element anchor, page URL, page index	The page URL and the annotation on the screenshot
Environment	Viewport size and device type (desktop, tablet, mobile) — not the OS, browser or handset model	Browser and OS
Console logs and network requests	No	Yes
Session replay	No	Yes, with data masking
Rewritten into an implementation prompt	Yes, one prompt per project	No
MCP for coding agents	Yes, 7 tools, OAuth 2.1, every account	Yes, generally available, every plan, no extra cost
Where it goes next	AI prompt, export, or MCP	Jira, GitHub, GitLab, Azure DevOps, Linear, Asana, ClickUp, Trello, Notion, Slack and more

Competitor details checked on 2026-09-01. Pricing and features change — check their site before deciding.

Who is doing the reporting

This is the fork in the road, and it decides everything after it.

Marker.io's reporter role is built for testers and end users. They submit through the widget and follow their own issues with no account, and that path is unlimited.
The richer client-collaboration role — guests with portal access — is a counted, capped seat on every tier. Count your client-side stakeholders before you pick one; the table below has the numbers and the date we checked them.
Writhink has no role to count. Someone opening the review link gives a display name and starts pinning; a token in their browser proves the comments are theirs, so they can edit, delete and reply without an account.
Page views are metered on Marker.io as well, which is an unusual second axis of cost when the widget sits on a busy production site. Writhink does not meter page views.
The trade is real. Writhink asks less of your client and captures less about their browser: no console logs, no network requests, no replay, and no browser or OS string at all.

What Marker.io meters, and what Writhink does not

	Writhink	Marker.io
Entry plan	Free during the open beta, no card	Starter $39/mo billed annually or $59 monthly: 3 users, 10 projects, 10 guests, 5,000 page views a month
Next tier up	No tiers during the beta	Team $149/mo annually or $199 monthly: 15 users, 50 projects, 50 guests, 25,000 page views a month
Agency plan	Not applicable	$129/mo, or $99 with yearly billing, for eligible small agencies: 15 members, 50 projects, 50 guests
Top tier	None	Business, custom pricing, yearly billing only
Trial	Not needed; it is free during the beta	15 days, no credit card
Guest reviewers	No account, not counted	Reporters unlimited; guests with portal access capped per tier
Page views	Not metered	Metered per tier

Competitor details checked on 2026-09-01. Pricing and features change — check their site before deciding.

A vague comment, step by step

Here is what happens to a sentence no tracker would accept as a ticket.

Step 1
The client writes it
"this bit is weird on my phone", pinned to the pricing table on /pricing, at a 390-pixel viewport, device recorded as mobile, with an automatic screenshot of the spot. Not which phone and not which browser — Writhink records the device type and the viewport, and nothing that identifies the handset.
Step 2
Writhink keeps the context, not just the words
Position, element anchor, page URL, page index, viewport size, device type and screenshot stay attached to that comment for as long as it exists, and follow it into everything downstream.
Step 3
The AI rewrites it without inventing anything
The item becomes a clear instruction naming the pricing table on the pricing page at a mobile breakpoint, with the screenshot beside it. What the client did not say is not filled in for them.
Step 4
Ambiguity is flagged, not resolved
If "weird" cannot be tied to anything specific, the item is marked "(unclear — ask the client)" and the prompt asks your AI to draft the question you send back.
Step 5
You paste it, or the agent fetches it
One prompt covers every open comment on the project. Or connect the MCP server with claude mcp add --transport http writhink https://mcp.writhink.com/mcp and ask Claude Code to read the feedback, set statuses and reply to the client itself.

Where each one belongs

Writhink fits when

The feedback comes from a client describing how something looks, not a tester describing how something broke.
An AI assistant does the implementation and you want the input written for it rather than for a developer's eyes.
You do not want to count guest seats, projects or page views.
Follow-up belongs in the comment thread the client already opened, rather than in a portal they have to be provisioned for.
You want to try it with no card and no tier decision.

Marker.io fits when

The report has to be diagnosable. Console logs, network requests and session replay with data masking are exactly what a hard bug needs, and Writhink has none of them.
The work has to land in an existing tracker. Writhink integrates with nothing.
You need to install without touching the code: a WordPress plugin or a browser extension across all browsers, desktop and mobile. Writhink's extension is Chrome only.
You want the agent to have the console output before it attempts a fix. Marker.io's MCP payload is richer than Writhink's for that job, and both servers are available on every account, so MCP itself decides nothing here.
You are running QA on a release with a real testing team, or you need a mature product with a long track record. Writhink is new and does not have one.

The cost shapes are different too

Marker.io's bill is a function of four things at once: internal users, projects, capped guest seats and page views. That last axis is the one people miss, because it couples your cost to your client's traffic rather than to your own usage, and a widget left on a busy production site is where the surprise arrives. The table above has the published figures and the date we checked them.

Writhink counts none of that today. It is free during the beta, with no seat, guest, project or page-view metering. That is also the weakness: a published, stable number you can put in a budget is something Marker.io has and Writhink does not, and if predictability matters more to you than the AI layer, that alone is a reasonable way to decide.

Before you count on it

Writhink is in open beta: everything described here is available and free, no credit card. Paid plans are announced but not purchasable yet. The one limit that applies today is five AI prompt generations a month per organization.

FAQ

Questions people actually ask

Does Marker.io have an MCP server?: Yes, and it is a good one. It is generally available, included on every plan at no extra cost, and gives an agent the full bug payload including console logs and network requests, so the agent can investigate, comment, attempt a fix and open a pull request. Claude Desktop, Claude Code, Cursor, Windsurf and any MCP-compatible client are supported.
Then why would I use Writhink's MCP instead?: Because the input is different, not because the server is better — having an MCP server is table stakes in this category now. Marker.io's payload describes a technical failure. Writhink's describes what a non-technical client wants changed, with the pin, the element, the breakpoint and the screenshot, and the prompt layer rewrites, merges and prioritizes those comments before any agent sees them. If your feedback is bugs, Marker.io. If it is client opinions, Writhink.
Do clients need an account with Marker.io?: Not for the reporter path — customers and end users submit through the widget and follow their own issues with no account. Guests, the richer collaborator role with portal access, are a counted and capped seat on each tier; the table above has the counts and the date we checked them. Writhink has no equivalent count; anyone with the link comments with a display name.
Can Writhink send feedback to Jira?: No. There are no native integrations of any kind. What you get instead is the generated prompt — copy it or download the .md — and the MCP server. An agent connected over MCP can read the feedback and act on it, but it will not create a Jira issue for you.
Which should I use for a release QA pass?: Marker.io. A QA pass produces reports that need reproduction steps, console output and environment detail, and it should land in your tracker. Writhink is for the client round before or after that, where the input is language rather than stack traces.

Start

Send one link. Get feedback you can build from.

Your clients comment right on the page — no account, no screenshots, no voice notes. Writhink sorts the comments and writes the prompt for your AI.