MCP server providing access to Webflow's Data API for managing sites, collections, pages, components, scripts, and AI chat
Critical gaps across all dimensions. Tool descriptions are present but generic (avg ~50 chars, well below 194-char baseline). Parameter descriptions exist in schemas but are minimal. No input schemas visible in source code for most tools, only Zod schema exports shown, not actual tool registration. Output schemas undocumented. Error handling absent. No tool annotations (readOnlyHint/destructiveHint). Composition is reasonable (29 focused tools), but definition quality is significantly below production baseline.
No input schemas visible in source code. Tool registration in src/tools/*.ts files not shown; only Zod schema exports visible. Cannot verify parameter types, constraints, or descriptions at registration point.
Tool descriptions are generic and minimal (~40 chars avg). Examples: 'ask_webflow_ai' lacks context on when to use it vs direct API calls. Descriptions do not explain WHAT the tool does, WHEN to use it, or what it returns. Below 194-char production baseline.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Risk levels are documented in the server spec (READ_ONLY, WRITE, DESTRUCTIVE) but not exposed to the MCP client. LLMs cannot distinguish safe vs dangerous operations without explicit hints.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 39 | 2026-07-28+ | v2 |
No error handling guidance. No recovery hints, retryability classification, or actionable error messages visible. LLMs will not know what to do if a tool call fails.
Output schemas undocumented. No visible documentation of what fields each tool returns, their types, or how to chain results to downstream tools. Agents cannot plan multi-step workflows.