Loading ToolBench
MCP server for managing Mlytics CDN infrastructure, including site management, DNS records, performance monitoring, cost analysis, and multi-CDN strategy optimization
Server has 6 tools with schemas and descriptions present, but quality is inconsistent. Tool naming follows verb_noun convention (create-, add-, update-, get-, automated-). However, descriptions are generic (avg ~80 chars, well below 194-char baseline). Critical gaps: (1) No output schemas documented for any tool, LLMs cannot plan downstream calls. (2) Parameters lack depth, many missing format/range constraints. (3) 'automated-login' exposes credentials as parameters (email/password), violating secret-injection pattern. (4) Error handling is absent, no recovery guidance. (5) Tool composition issues: 'update-domain' and 'update-strategy' are vague about what fields actually change. Schemas are present but minimal; descriptions are present but underdeveloped.
Add a DNS record to a site with name, type, TTL, and values
Automated login to Mlytics Portal using email and password credentials with optional headless browser mode
Create a new CDN site with domain, name, description, and optional CDN providers
Retrieve performance metrics for CDN providers across regions with optional time range and provider/region filters
Update domain settings including CDN cache, security, and WAF configuration
Update CDN strategy with type (geo, performance, cost, availability, hybrid) and configuration weights
Credentials exposed as tool parameters: 'automated-login' accepts email and password directly. Credentials must never appear in tool parameters, use server-side secret injection via environment variables or vault. Agent traces log all parameters, leaking secrets into logs and prompt history.
No output schemas documented for any tool. LLMs cannot determine what fields to expect, preventing downstream tool chaining and forcing wasteful discovery calls. Every tool must document its return structure.
Descriptions are generic and lack actionable context. Average ~50 chars vs. 194-char baseline. Descriptions do not explain WHEN to use each tool, what it modifies, or what to do next. E.g., 'Update domain settings' does not clarify which settings or whether this is idempotent.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 54 | 2026-07-28+ | v2 |
No error handling or recovery guidance. Tools do not return actionable error messages. If a DNS record add fails, the LLM has no guidance on what to try next. Errors must categorize as retryable, user-fixable, or fatal.
Parameter constraints are incomplete. Numeric parameters (ttl, cache_ttl, weights) lack min/max bounds. Enum values are present for some (dns type, strategy type) but descriptions do not explain the implications of each choice. Format constraints (e.g., domain name validation) are absent.