Mimiq: test pages and sign-up flows on simulated people

Compare two versions of copy

mimiq.compare_copy

Compare two versions of copy or an email on the same simulated people and get Mimiq's call on which to ship. The same people see version A and version B. Mimiq's call combines a forecast of how people like them behave, asked in both orders, with how the same people moved: clear, leaning, or too close to call (no pick). Returns the call, the reason in one sentence, how many people moved toward each version, each version's top objections, and a link to the report. Without a key, an agent also gets one free A/B test on up to 10 simulated people per version, while the daily free A/B allowance lasts. With a key, any size. The forecast has held up on headlines (76% right on 1,000 real headline A/B tests, against 61% for the best rule of thumb); on emails, text messages and ads it was no better than chance at ranking small wording changes, and on real web page tests it did not beat the simplest guess.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`count`	No	How many simulated people see both versions (default 10). Use 5 or more: from 5 people up, Mimiq reads who moved. Uses count credits to recruit plus count for version B.
`format`	No	"copy" for a headline, tagline or short message people come across; "email" for an email in an inbox (a first line "Subject: ..." is read as the subject).	copy
`audience`	No	Who the simulated people should be, in plain words, e.g. "startup founders in SF" or "parents shopping for kids toys". If omitted, Mimiq picks the likely audience from the content.
`version_a`	Yes	Version A, usually the current one.
`version_b`	Yes	Version B, usually the new one.
`timeout_seconds`	No

Output Schema

TableJSON Schema

Name	Required	Description	Default
`run`	No
`call`	No
`tool`	Yes
`status`	No
`report_url`	No
`summary_card`	No

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedSep 30, 2026

changedOutput schema / (root)

Previous value: -nullNew value: +{
+  "additionalProperties": true,
+  "properties": {
+    "call": {
+      "additionalProperties": true,
+      "properties": {
+        "headline": {
+          "type": "string"
+        },
+        "pick": {
+          "type": [
+            "string",
+            "null"
+          ]
+        },
+        "reason": {
+          "type": "string"
+        }
+      },
+      "type": "object"
+    },
+    "report_url": {
+      "type": [
+        "string",
+        "null"
+      ]
+    },
+    "run": {
+      "additionalProperties": true,
+      "type": "object"
+    },
+    "status": {
+      "type": "string"
+    },
+    "summary_card": {
+      "type": "string"
+    },
+    "tool": {
+      "type": "string"
+    }
+  },
+  "required": [
+    "tool"
+  ],
+  "type": "object"
+}

AddedSep 24, 2026

TDQS

A4/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With readOnlyHint=false already signalling a non-read operation, the description meaningfully extends it: it discloses credit consumption ('Uses count credits to recruit'), the keyless free allowance (10 people per version), and unusually honest accuracy caveats (76% on headlines, no better than chance on emails/ads, no edge on web pages). This is exactly the behavioral context an agent needs beyond the annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose, mechanism, return payload, pricing, and caveats are front-loaded in that order, which reads well. The final accuracy sentence is dense and long, but it earns its place as material disclosure rather than filler, keeping the overall structure efficient.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a moderately complex, cost-incurring comparison tool with an output schema, the description covers mechanism, cost, free-tier limits, output shape, and reliability caveats. Nothing an agent needs to invoke this correctly and set expectations is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is high at 83%, and the schema already documents count, format, and audience in detail, so the schema does the heavy lifting. The description reinforces the mechanics of A/B comparing the same audience and the count behavior, but adds little syntactic meaning beyond the schema. Baseline 3 is appropriate.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description states a specific verb+resource ('Compare two versions of copy or an email on the same simulated people') and clarifies the mechanism (same people see version A and B). It implicitly separates itself from mimiq.compare_urls via the copy/email scope, but never names a sibling to sharpen the boundary with test_copy or test_text.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Usage context is implied and supported: it explains the format toggle (copy vs email), the free-tier A/B allowance, and the credit cost of count. However, there is no explicit when-to-use-this-vs-alternative guidance, e.g. how it differs from test_copy or compare_urls, leaving the agent to infer the choice.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Mimiq: test pages and sign-up flows on simulated people tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.