refero-design-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| REFERO_VERBOSE | No | Diagnostics to stderr | false |
| REFERO_SITE_URL | No | Origin base URL | https://styles.refero.design |
| REFERO_CACHE_DIR | No | On-disk cache location | ~/.cache/refero-design-mcp |
| REFERO_TIMEOUT_MS | No | Per-request timeout | 30000 |
| REFERO_USER_AGENT | No | Request identification | see src/config.ts |
| REFERO_CONCURRENCY | No | Max simultaneous requests | 4 |
| REFERO_MAX_FETCHES | No | Max page fetches per tool call | 25 |
| REFERO_CACHE_TTL_MS | No | Cache freshness window (7 days) | 604800000 |
| REFERO_STYLES_SITEMAP | No | Style index | {site}/sitemaps/styles.xml |
| REFERO_MAX_RESPONSE_CHARS | No | Response budget | 40000 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| resources | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| refero_index_statusA | Report how many published styles exist and how many are indexed locally. Call this before concluding that a style does not exist: a miss usually means it has not been indexed yet. |
| refero_search_stylesA | Find design styles on styles.refero.design by free text. Matches style names, north-star statements, colours, fonts and URLs on word boundaries. The local index grows on demand, so results improve as more styles are indexed. |
| refero_match_styleA | Given a design brief in prose, return the styles that best match it, each with the terms that triggered the match and a short rationale. Use this when the request is qualitative ("I need a dense, data-heavy dashboard") rather than a specific name. |
| refero_get_design_mdA | Render a style as a design.md document. The |
| refero_get_styleA | Return the parsed design system plus the measured tokens scraped from the live site (colours, fonts, radii, spacing) as JSON. Prefer |
| refero_list_style_idsA | Return published style UUIDs and their last-modified timestamps, straight from the public sitemap. Use this to discover ids without reading any style page. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 6 tools
Search vs match overlap somewhat (both find styles), but descriptions clearly separate keyword matching from qualitative brief matching. The two retrieval tools (get_design_md for prose, get_style for JSON tokens) are explicitly differentiated, and index_status/list_style_ids target distinct metadata concerns.
Every tool follows the identical refero_<verb>_<noun> pattern: refero_search_styles, refero_match_style, refero_get_style, refero_list_style_ids, etc. No camelCase or naming deviations; totally predictable.
Six tools is well-scoped for a design-style discovery and retrieval server, with each tool earning its place across discovery, matching, and rendering. No redundancy or bloat.
Covers the full read lifecycle: discover ids, search, brief-match, fetch JSON tokens, and render prose. Being a read-only scraper of a public site, no CRUD is needed, though minor gaps (e.g. pagination/batch retrieval) exist.