astrondesign
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| ASTRONDESIGN_PYTHON | No | The python executable used for the UI UX Pro Max search (design_search). | python |
| ASTRONDESIGN_BROWSER | No | Browser executable path (highest priority when launching the browser for design_render/design_inspect). Optional; falls back to Playwright default and then Edge. | |
| ASTRONDESIGN_IDLE_MS | No | Time in milliseconds after the last call before the browser is closed. | 120000 |
| ASTRONDESIGN_NETWORK | No | External network access is blocked by default. External requests are only allowed when this is set to `allow` AND the tool call argument `allow_network:true` is also given. | off |
| ASTRONDESIGN_OUT_DIR | No | Output root for artifacts (screenshots, report.json, Design Feedback); must be an absolute path. | <cwd>/.astron-logs/astrondesign |
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
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| design_guideA | UI/UX design harness guide: workflow, conflict precedence, critique protocol and references (impeccable, taste, UI UX Pro Max, brand DESIGN.md). Call first without topic. |
| design_searchC | Search UI UX Pro Max design data (styles, palettes, fonts, UX rules) or build a design-system suggestion. Results are recommendations, not rules. |
| design_directionC | Deal an art direction for a public landing or marketing page (taste-skill draw). Not for dashboards or app screens. |
| design_lintC | Static UI design lint of source files with impeccable and taste rules, resolved by AstronDesign precedence. Writes nothing to the project. |
| design_renderC | Render a page in a real headless browser at several widths: screenshots, measured design rules, accessibility snapshot. External network blocked by default. |
| design_inspectC | Inspect page elements by selector or point (HTML, 16 computed styles, bounds, crop) and write a Design Feedback note. |
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
Each tool has a distinct role: guidance (design_guide), static analysis (design_lint), data lookup (design_search), ideation (design_direction), browser rendering (design_render), and element-level inspection (design_inspect). The only real friction is between design_render and design_inspect, both of which measure live pages, though their descriptions (whole-page screenshots/rules vs. single-element computed styles) differentiate them adequately for a careful agent.
All tools share the design_ prefix with a single word suffix, forming a predictable namespace. The second token mixes nouns (guide, direction) and verbs (lint, render, inspect, search), so the verb_noun pattern is not strictly followed, but the prefix uniformity keeps the set readable and scannable.
Six tools is well within the ideal range and each one covers a separate phase of a design workflow (guide, search, direction, lint, render, inspect). No tool appears redundant or tacked on.
The surface covers the full read/analyze/advisory loop: onboarding guidance, reference lookup, direction dealing, static linting, rendering, and element inspection with feedback notes. The notable gap is any apply/fix or note-retrieval operation — design_lint explicitly writes nothing and there is no tool to read back or resolve accumulated Design Feedback notes, which could dead-end a remediation workflow.