DiffiCat MCP server

• Registry namecom.difficat/difficatverified in the MCP registry

The registry name sits under difficat.com. The registry only lets a publisher use that namespace after proving control of the domain through DNS or HTTP.

• Stateanswerschecked 2026-10-24
• Tools15
• Server calls itselfdifficat
• Server version3.129.1
• Protocol version2025-06-18

On 2026-10-24 the server answered initialize and tools/list without a sign-in. First answer in 222 ms.

• Its tools do every use case we checked: compare old and new site pages for missing or broken URLs; check page layout differences at phone, tablet and desktop widths.
• Checked on 4 of the 7 days it takes to judge reliability.

Said by rules from the measurements above. The likely first call is a model's pick from the tool list; read its description before relying on it.

• INITE MCP scoreafter a weeka score needs 7 daily checks: 4 so far
• What an agent can do with it2 of 2 use cases
• Answeredmeasuring4 of the 7 daily checks it takes to print a share
• Median first answer422 ms
• Space in an agent's context≈ 7,052 tokensthe definitions of all its tools (15), which an agent carries on every turn
• Tool definitionsA · 99 of 100
• Typed results15 of 15 toolsdeclared outputSchema: results come typed, read once

• Parameters without a type: in 1 of 15

A program checks the server once a day: initialize and tools/list, no model. Tokens are the definitions' length at about four characters a token. The grade averages six checks of every tool's name, description, schema and annotations; it says how well the tools are described, not whether they work. The score weighs what was measured and nothing else: answering (25%), doing the product's job (25%), definitions (20%), context (10%), use (10%) and how readily a client connects (10%). A part that was not measured is left out, and a hidden instruction caps the score at 40. The state of vendors' MCP servers: method and numbers →

• discover_pages

  Free 5 times a day per client address (at most 50 pages each), no key needed. Lists the pages of Site A (sitemap, robots.txt or crawl), mapped onto Site B when given. Call it to see which pages a comparison would cover or to pick pairs…

• capture_screenshot

  Free 5 times a day per client address, no key needed. Full-page PNG of one URL at the given viewport width, returned inline when small, else as links.file to fetch with the jobToken. Call it for a one-off look at a page or a mobile-width…

• start_comparison

  Free up to 40 pages a day per client address, no key needed; sample report: https://difficat.com/agents/sample/report.json. Compares Site A against Site B on page existence, content and visual regression, priced per page compared. For a…

• recover_comparison

  Returns the job or screenshot the payment credential in _meta paid for, with a new jobToken that replaces the old one; or, when the request carries a key and _meta no credential, the latest job that key paid for (no jobToken: the key…

• get_comparison

  Status and progress of a job and, once it is terminal, its summary: passed, failed and issuesBySeverity. To wait for the end, call wait_for_comparison instead. Free.

• cancel_comparison

  Stop a running job; the job reports cancelled within a minute, with the pages compared so far in its report. Free.

• wait_for_comparison

  Waits until the job reaches a terminal status (completed, failed or cancelled) or timeoutSeconds pass (default 60, at most 300), then answers like get_comparison: status, progress and, once terminal, the summary with passed, failed and…

• get_report

  The finished report. Call it first with failedOnly: true and view: "digest": results, total and nextOffset then cover only the pages that failed, one small object per page with its issues, diff percentage and largest visual regions, and a…

• get_screenshot

  Fetch one screenshot PNG by filename. Call it when a get_report page result or a capture_screenshot result names a screenshot you want to see; a PNG over the inline cap comes back as links.file to fetch with the jobToken. Pass a report…

• get_balance

  The prepaid balance and the last 20 ledger entries of the key the request carries in its Authorization: Bearer or X-DiffiCat-Key header. Call it to check what a key can still pay for; request_prepaid_key is the way to obtain a key. The…

• request_prepaid_key

  Mints a prepaid key and a Checkout link for a human to pay by card, so later paid calls are debited from its balance instead of challenged. Call it when a call was refused for payment and your human prefers a card to a wallet; with a key…

• send_feedback

  Sends one message to the people who run this service: why a price, a payment path or a result did not fit. Call it when you stop short of a paid call or a result was not what the task needed; a person reads every message. Do not call it…

• list_snapshots

  Lists the saved snapshots the DiffiCat account behind the request’s key owns or was given access to, newest first, each with access owner or reader. Call it to find a snapshotId for get_snapshot or compare_snapshot. Needs an account’s key…

• get_snapshot

  A saved snapshot of a site: without n, the snapshot (label, origin, width, check mode, status) with its page list (n, path, status, whether a screenshot exists) and your compares that use it; with n, that page’s stored record: path…

• compare_snapshot

  Free up to 40 pages a day per client address for a live compare, no key needed. Compares a saved snapshot, as Site A, against a live targetUrl (every stored page paired by path, or the urls pairs you name) at the snapshot’s width and…