Website QA testing tool
Most bug reports arrive missing the one thing that would have made them cheap to fix. The reporter saw a console error and did not know to look; the network request failed silently; the screenshot shows the symptom but not the breakpoint. The developer’s first hour goes on reconstructing what the reporter’s browser already knew.
A QA tool earns its cost by removing that hour. The question is not whether it lets someone annotate a page — they all do — but how much of the failure state it captures without the reporter doing anything, and what happens to that data afterwards.
What gets captured on each report
Layerdock runs as a browser extension. The reporter clicks the broken element and types what they saw. Attached automatically, with no action from them:
- The exact DOM element and a generated CSS selector, plus a screenshot of it in context.
- The last 50 console entries with stack traces — errors, warnings and logs from before the report, not just after it.
- The last 50 network requests with URL, method, status and duration, so a 500 or a 4-second call is visible without reproducing anything.
- Browser, OS, screen resolution, viewport size, device pixel ratio, language and hardware details — the environment fields a bug template asks for and reporters guess at.
- The page URL and the viewport the reporter was in when they hit it.
The console and network bundle is on the $49/mo Agency plan, not the $19 Pro plan. Capture itself is never gated: the extension records the bundle on every pin regardless of plan, so upgrading reveals the history you already collected instead of starting from empty.
Handing the bug to an AI coding agent
This is the part with no equivalent in the tools below. Layerdock exposes feedback over the Model Context Protocol, so Claude Code, Cursor or any MCP client reads the report directly — selector, page, screenshot, and on Agency the console and network bundle. The agent starts from the actual failure state rather than from a description someone retyped into a chat window.
The full MCP integration exposes five tools, including a fix brief the agent can grep your source with. On the Pro plan it carries element context and screenshots but not the diagnostics bundle. If handing full failure data to an agent is the reason you are evaluating this, that is an Agency-plan capability.
Session-capture tools are a different category
There is a separate class of product built for engineers reporting bugs to other engineers: you hit record, it captures the session with full console and network data, and you send a link. Those tools capture more raw detail than any review tool does, and if everyone finding your bugs is a developer, that is the shape you want. The full survey covers them.
What they do not have is a workflow for a non-technical reporter — no client review round, no stakeholder sign-off, no share link you can hand someone outside the team. The moment a project manager, a client or a QA tester without a developer’s instincts is the one finding the bug, that gap is the whole problem, and it is what a live-site review tool exists to close.
Against the review tools
Among tools a non-technical reporter can actually use, BugHerd ($50/mo for 5 members) attaches browser and OS metadata but not full console output. Marker.io ($39/mo annually for 3 users) does capture console, network and session replay — on its Team plan at $149/mo, and it installs as a script tag, so it cannot review an environment you have no deploy access to. Cheaper entry-tier options with console capture do exist in this category — the full roundup names them and their prices.
Layerdock sits at $19/mo Pro or $49/mo Agency per workspace, extension-based so it works on any URL, with the diagnostics bundle and the MCP integration on Agency.
Common questions
What is the best QA tool for website bug reporting?
It depends who is finding the bugs. If they are all engineers, a session-capture tool aimed at developers will record more than any review tool does. If they include clients, testers or stakeholders, you need a live-site review tool the reporter can use without training — Layerdock, BugHerd or Marker.io. Layerdock captures the element, a screenshot, console errors and failed network requests on every report, on the $49/mo Agency plan.
Does Layerdock capture console errors and network requests?
Yes, on every pin, on the $49/mo Agency plan. The bundle holds the last 50 console entries with stack traces and the last 50 network requests with URL, method, status and duration, plus browser, OS, screen resolution, viewport and device details. Capture itself is never gated — the extension records the bundle on every pin whatever your plan — so upgrading reveals the full history rather than starting from empty.
Can I use it on staging and production?
Yes. Layerdock is a browser extension rather than a script tag, so it works on any URL without a code change — staging, production, or a client environment you have no deploy access to. Widget-based tools like Marker.io and Feedbucket need the tag deployed to each environment first.
How does the MCP integration work?
Layerdock exposes feedback over the Model Context Protocol, so an AI coding agent — Claude Code, Cursor, or any MCP client — can read a bug report directly: the element selector, the page URL, the screenshot, and on the Agency plan the console and network bundle. The agent works from the actual failure data instead of a pasted description. No other tool in our comparison set offers this.
Does it replace automated testing?
No. Layerdock is for human-found bugs — exploratory testing, UAT, client review, stakeholder sign-off. It does not run test suites and does not replace Playwright, Cypress or unit tests. It is the layer that turns "this is broken" from a human into a ticket a developer can act on.
Read how to report a bug with screenshot and console errors, connect your coding agent over MCP, compare 17 feedback tools, or check pricing. Layerdock is pre-launch — the pricing page takes waitlist signups rather than payment.