MCP server for state-of-the-art CSS: Baseline status, browser support, and CSS source audits.
Server has 6 tools with clear verb-noun naming (search_, get_, check_, audit_) and reasonable descriptions (avg ~80 chars). All tools are READ_ONLY and stateless. However, parameter descriptions are sparse or missing entirely. Input schemas exist but lack depth: most parameters are bare strings with minimal constraint documentation. Output schemas are not documented in the visible code. Error handling converts upstream failures to tool results (good), but recovery guidance is minimal. No tool annotations (readOnlyHint, etc.) despite all being read-only. Composition is sound, tools are single-purpose and chainable.
Audit a CSS stylesheet for browser support against a target baseline
Check which browser versions support a CSS property, value, or selector
Get design guidance or page quality assessment
Get everything about one CSS feature
Search for CSS features by name, id, or description
Find CSS features that became usable recently
Output schemas not documented. Tools return results but LLMs cannot predict field structure, forcing trial-and-error parsing and wasting tokens.
Parameter descriptions are minimal or absent. 'query' in search_css_features lacks guidance on format/scope. 'target' in audit_css lists enum values but no explanation of when to use each baseline level.
Tool 'dont_make_me_think' has vague naming and description. Does it critique CSS, assess UX, or provide design advice? The name is colloquial and does not follow verb_noun convention. Ambiguous purpose will cause LLMs to misselect it.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, idempotentHint) despite all 6 tools being read-only and safe to retry. Agents cannot infer safety properties and may avoid retrying on transient failures.
Error handling converts failures to tool results (good), but recovery guidance is generic. 'Upstream lookup failed' does not tell LLM whether to retry, fall back to check_support, or ask the user.