Melytics MCP has 9 well-named, read-only analytics tools with consistent parameter schemas and clear descriptions. All tools follow verb_noun naming (list_sites, get_stats, get_breakdown, etc.) and include descriptions (avg ~100 chars). Input schemas are present and typed with Zod. However, output schemas are not documented, responses are JSON-stringified without field-level documentation. Parameter descriptions are adequate but lack format constraints (e.g., date format YYYY-MM-DD is stated but not enforced via regex). No error handling guidance, no pagination support despite tools returning potentially large result sets (e.g., get_breakdown with limit up to 100). All tools are read-only (low risk), but composition could be improved, no guidance on which tools to chain together.
Chart annotations (deploys, launches, campaigns) noted on the dashboard
Top values for one dimension (pages, referrers, countries, devices, browsers, OS, UTM, events)
Break down a custom event by one of its properties. A string property returns a value distribution; a numeric property returns an aggregate (sum/avg/count); a numeric property with "by" set is summed per group (e.g. purchase value by product = revenue by product)
Funnel step-by-step visitor counts and drop-off rates
Goal conversion counts and rates for a site
Visitors online right now (last 5 minutes) and the pages they are on
Output schemas not documented. Tools return JSON-stringified responses without field-level documentation. LLMs cannot plan downstream operations or extract specific fields reliably.
No pagination support. get_breakdown accepts limit up to 100 but no offset/cursor or total_count. Large result sets risk context window exhaustion.
Date format constraints stated in descriptions but not enforced via regex or format validators. LLMs may pass invalid dates like '2024-13-45'.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2026-07-28+ | v2 |
Traffic overview for a site: time series, totals, and previous-period comparison
p75 Core Web Vitals (LCP, CLS, INP, TTFB) for a site
List the sites tracked by this melytics instance
No error handling guidance. If API returns 4xx/5xx, error message is generic ('melytics API 401 for /sites'). LLM cannot determine if error is retryable or user-fixable.
No tool composition guidance. Descriptions do not indicate which tools to call first (e.g., 'Call list_sites first to get site_id') or how results chain together.