Is the Difficat MCP server down?

mcp-registry/com.difficat/difficat · v3.120.0

healthy
Last checked 2026-10-23 05:30 UTC · handshake 215ms

As of 2026-10-23 05:30 UTC, the Difficat MCP server is up with a Glimind reliability score of 93/100 (grade A). Over the last 30 days it answered 100% of Glimind's liveness checks. It had 1 outage in 90 days, with a median recovery of 6.1 hours.

uptime (30d checks)
100%
checks observed
4,131
outages (90d)
1
tools
15

Compare a website before and after a migration: missing pages, content changes, visual diffs.

Install Difficat MCP server

Add to Cursor Add to VS Code

Claude Code

claude mcp add --transport http difficat-mcp-server https://app.difficat.com/mcp

Claude Desktop · Windsurf · Cline (mcpServers JSON)

{
  "mcpServers": {
    "difficat-mcp-server": {
      "type": "http",
      "url": "https://app.difficat.com/mcp"
    }
  }
}

Third-party server, listed as discovered. Review its source before installing.

Agent-readiness: A (92/100)

A (92/100): 0 error(s), 2 warning(s) across 15 tool(s); ~6,794 tokens of tool definitions.
Scores higher than 74% of 15,586 MCP servers Glimind has audited.
Its tool list costs an agent ~6,794 tokens per session — more than 65% of servers.
91 → 92 since 2026-10-17
no tool-poisoning patterns
Schemas87
Descriptions100
Safety hints100
Token cost87
Client compatibility100
Protocol100
Reliability100
Stability20

What to fix (3)

1 tool definition(s) exceed ~1,200 tokens each (largest: start_comparison, ~1,707).

Move long reference material out of descriptions (link a doc or expose a resource), drop repeated examples, and collapse large enums. Trimming these to ~600 tokens saves ~1,107 tokens every session. How to fix →

Tools: start_comparison · saves ~1,107 tokens/session

1 tool(s) have parameters with no type (no type/enum/anyOf/$ref).

Give every parameter a "type" (or "enum"/"anyOf"). Untyped parameters make models guess — strings for numbers, objects for arrays. How to fix →

Tools: compare_snapshot

In the last 30 days: 0 tool(s) removed and 41 schema change(s).

Add new tools or optional parameters instead of renaming or removing; deprecate first (say so in the description) and remove later. Agents cache tool contracts. How to fix →

Auto-fix: 1 corrected tool definition

Paste-ready definitions with the problems above fixed. Changes marked verify need a human look. All of them as JSON: /v1/readiness/mcp-registry%2Fcom.difficat%2Fdifficat/fixes

compare_snapshot 1 change
{
  "name": "compare_snapshot",
  "description": "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 check mode, priced at one captured page per pair (half a live pair) with no discovery; the snapshot’s own origin is never fetched. With otherSnapshotId instead of targetUrl it compares two snapshots captured alike at no charge, which needs the account’s key on the r",
  "inputSchema": {
    "type": "object",
    "$schema": "http://json-schema.org/draft-07/schema#",
    "required": [
      "snapshotId"
    ],
    "properties": {
      "urls": {
        "type": "array",
        "items": {
          "type": "object",
          "required": [
            "source",
            "target"
          ],
          "properties": {
            "source": {
              "type": "string",
              "maxLength": 2048,
              "minLength": 1,
              "description": "Page URL on Site A."
            },
            "target": {
              "type": "string",
              "maxLength": 2048,
              "minLength": 1,
              "description": "The same page's URL on Site B."
            }
          }
        },
        "maxItems": 50000,
        "minItems": 1,
        "description": "With targetUrl: the pairs to compare, each source a page of the snapshot (its path or full URL on the snapshot’s origin) and each target on targetUrl. Without it every stored page is paired by path."
      },
      "dryRun": {
        "type": "boolean",
        "description": "Quote only: runs every check and the pricing, then answers the page count and exact amount. Nothing is charged, the free allowance is not used and no job starts; the same body without dryRun is priced from the retained result."
      },
      "visual": {
        "description": "Not accepted: a run against a snapshot uses the masks and freeze recorded with the snapshot.",
        "type": "string"
      },
      "targetUrl": {
        "type": "string",
        "maxLength": 2048,
        "minLength": 1,
        "description": "The live site to compare against the snapshot (full URL). Give this or otherSnapshotId."
      },
      "threshold": {
        "type": "number",
        "maximum": 1,
        "minimum": 0,
        "description": "Visual diff tolerance as a fraction from 0 (exact) to 1: 0.05 means 5%. The web app takes a percentage; the API does not."
      },
      "shareToken": {
        "type": "string",
        "maxLength": 200,
        "minLength": 20,
        "description": "A share token (dcs_…) the snapshot’s owner issued for agents. Needed unless the account key on the request’s HTTP headers owns or was granted the snapshot."
      },
      "snapshotId": {
        "type": "string",
        "pattern": "^snap-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$",
        "description": "The snapshot id, snap-<uuid>, from list_snapshots or the web app."
      },
      "callbackUrl": {
        "type": "string",
        "maxLength": 2048,
        "minLength": 1,
        "description": "Optional https URL (port 443) that receives one JSON POST when the job ends (jobId, status, summary, links; never the jobToken), a nudge to read the job. Public hosts only, no redirects followed, retried twice; delivery never affects the job."
      },
      "maxPriceUsd": {
        "type": "number",
        "description": "Optional spend cap in USD. A job priced above it is refused before any challenge, so an auto-paying client is never asked for more.",
        "exclusiveMinimum": 0
      },
      "idempotencyKey": {
        "type": "string",
        "maxLength": 255,
        "minLength": 1,
        "description": "Optional opaque string a retry repeats: under a prepaid balance the same value within 24 hours is not charged again and answers DUPLICATE_REQUEST with the jobId of the first call."
      },
      "otherSnapshotId": {
        "type": "string",
        "pattern": "^snap-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$",
        "description": "A second snapshot, captured at the same width and check mode, to compare the first against. Needs the account key on the request; never free. Give this or targetUrl."
      }
    }
  },
  "annotations": {
    "readOnlyHint": false,
    "openWorldHint": true,
    "idempotentHint": false,
    "destructiveHint": false
  },
  "title": "Compare against a snapshot"
}

Get alerted

Get an email or webhook when this server's readiness drops by 10+ points, tool-poisoning text appears, or a tool breaks:

curl -X POST https://glimind.com/v1/subscriptions -H 'content-type: application/json' -d '{"tools":["mcp-registry/com.difficat/difficat"],"email":"[email protected]"}'

Agents: watch_tool({"toolId": "mcp-registry/com.difficat/difficat"}) over MCP.

Passed

Largest tool definitions

Static checks of the advertised tools/list (schema, descriptions, annotations, tool-poisoning patterns, names, size) plus observed protocol, uptime, latency and contract changes. Read-only: no tool is ever called. Weights: schema 15, docs 20, safety 15, efficiency 15, compatibility 15, protocol 5, reliability 10, stability 5. Token counts are estimates (~3.6 characters per token).

Operator? Fix these, then re-check live. Embed the readiness badge:

Markdown
[![MCP agent-readiness](https://glimind.com/v1/readiness/badge/mcp-registry%2Fcom.difficat%2Fdifficat.svg)](https://glimind.com/server/mcp-registry%2Fcom.difficat%2Fdifficat?ref=badge)
HTML
<a href="https://glimind.com/server/mcp-registry%2Fcom.difficat%2Fdifficat?ref=badge"><img src="https://glimind.com/v1/readiness/badge/mcp-registry%2Fcom.difficat%2Fdifficat.svg" alt="MCP agent-readiness"></a>
reStructuredText
.. image:: https://glimind.com/v1/readiness/badge/mcp-registry%2Fcom.difficat%2Fdifficat.svg
   :target: https://glimind.com/server/mcp-registry%2Fcom.difficat%2Fdifficat?ref=badge
   :alt: MCP agent-readiness

Tool-definition log

Glimind records every version of this server's tool definitions in a signed, append-only transparency log. Current manifest sha256:a8a971ec40bc0b3c…, first seen 2026-10-22, included at leaf 45689 of 50,239. Version history · verify what your client received.

Tools (15)

Compare: most reliable web servers · most reliable finance servers · most reliable files servers · most reliable other servers · most reliable payments servers

Know before your agent calls it

Glimind tells agents whether a server is up right now and how to call it, then fails over to a live alternative when it isn't.

Add Glimind — free, no signup:

Add to Cursor Add to VS Code

Claude Code

claude mcp add --transport http glimind https://glimind.com/mcp

Claude Desktop · Windsurf · Cline (mcpServers JSON)

{
  "mcpServers": {
    "glimind": {
      "type": "http",
      "url": "https://glimind.com/mcp"
    }
  }
}

FAQ

Is the Difficat MCP server down right now?

As of 2026-10-23 05:30 UTC, the Difficat MCP server is up with a Glimind reliability score of 93/100 (grade A). Over the last 30 days it answered 100% of Glimind's liveness checks. It had 1 outage in 90 days, with a median recovery of 6.1 hours.

How many tools does Difficat MCP server provide?

Difficat MCP server advertises 15 tools, including cancel_comparison, capture_screenshot, compare_snapshot, discover_pages, get_balance.

How do I install Difficat MCP server?

Point your MCP client at this streamable-http endpoint: https://app.difficat.com/mcp

Is the Difficat MCP server safe to connect to an AI agent?

Glimind scanned all 15 tool definitions for tool-poisoning patterns — hidden or invisible characters, instructions that override the model, concealment from the user, requests to read credential or MCP config files, conversation exfiltration and cross-tool shadowing — and found none. This checks what the server tells the model, not what its code does; review the source before granting access to sensitive data.

Operator of Difficat MCP server?

Status page: 90-day uptime, every outage and how long it lasted.

Claim this listing to get the Verified operator marker, a remediation checklist and owner analytics, or embed the live badge from any of the tool pages above. Maintainers: alerts, deploy attestation and monthly reports.

Status is an automated estimate from Glimind's liveness checks (handshake and tool listing only; no tool is ever called) and may lag or differ from your network. Glimind is not affiliated with this server or its operator. Operator? Request a correction or opt out at https://glimind.com/opt-out.

Feedback
Help us improve