Mimicus
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MIMICUS_BROWSER_EXECUTABLE | No | The absolute executable path of an installed compatible browser to use. Required on non-Windows operating systems, or to explicitly select a different installed compatible browser. |
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 |
|---|---|
| mimicus_searchB | Find current design references from the seven live source indexes. A new search fetches fresh data; operationId explicitly continues a bounded RAM-only operation. Native labels and computed similarity remain distinct. |
| mimicus_browse_categoriesC | Explore source-native taxonomy in an active operation, or fetch a fresh bounded source view. Normalized concepts are separately labeled. |
| mimicus_list_sourcesB | List all seven adapters, actual current-operation diagnostics, supported interfaces, access restrictions and temporary resource state. An unqueried descriptor is not a live connection. |
| mimicus_get_referenceB | Retrieve an active discovery candidate with source-native classifications, original URLs, provenance and source-supported remote preview links. No disk snapshot fallback. |
| mimicus_end_discoveryB | Release the selected temporary discovery operation and its candidates/cursors. |
| mimicus_open_referenceB | Open an authorized original site in an isolated temporary Chromium session and return its actual screenshot as MCP image content. Supply either a live reference pair or an explicit original URL; additional resource origins must be approved. |
| mimicus_capture_viewB | Return an actual current viewport or bounded selected-region screenshot directly as MCP image content; no persistent screenshot file is created. |
| mimicus_explore_pageC | Find current semantic sections, navigation and controls with geometry and selectors. This inspects only the selected open page. |
| mimicus_interactA | Safely explore a selected page: hover, focus, non-destructive click, scroll, pointer, same-origin reviewed navigation, keyboard or viewport/reduced-motion change. Returns the resulting real image. |
| mimicus_capture_motionB | Capture a selected effect as actual timestamped images, optionally applying one safe interaction between initial and subsequent frames. No automatic whole-site recording. |
| mimicus_close_sessionB | Close the browser/context/proxy and release temporary page and analysis state. |
| mimicus_analyze_elementB | Measure a selected component in the active page, including optional controlled interaction, scroll container and responsive viewports. Returns direct evidence without storing a pattern repository. |
| mimicus_get_design_insightsB | Return measured layout, palette/spacing/typography/navigation/animation observations and separately labeled original guidance for the selected live page, alongside its real visual reference. |
| mimicus_get_implementation_specA | Get actionable measured constraints and original implementation guidance from a selected temporary analysis. Requires an active session; there is no global stored pattern ID. |
| mimicus_compare_referencesB | Compare selected open references with actual images and measured insights. Supply a project brief to synthesize grounded original composition, navigation and motion guidance; no aesthetic winner is inferred. |
| mimicus_verify_implementationB | Reobserve an authorized open implementation and compare it with selected temporary reference evidence across matching viewports. Report differences/tolerances rather than claiming visual identity. |
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 16 tools
The verbose descriptions attempt to distinguish tools, but several analysis-oriented tools overlap conceptually: get_design_insights and get_implementation_spec both return measured layout/guidance, while analyze_element and explore_page both inspect page structure. Similarly capture_view, capture_motion, and open_reference all produce screenshots, which risks misselection.
All tools share the mimicus_ prefix and follow a mostly predictable verb_noun pattern (get_reference, capture_view, analyze_element, verify_implementation). The lone bare verb 'interact' is a minor deviation but the set is largely coherent.
16 tools is on the heavier side but justified by the multi-stage workflow (discovery, session capture, analysis, comparison, verification). Each tool maps to a plausible step, though a few analysis tools could be consolidated.
The surface covers the full lifecycle: discovery/search, browsing, opening, capture (view/motion), interaction, measurement, insights, spec generation, comparison, and implementation verification, with clean teardown tools (end_discovery, close_session). Minor gaps exist around explicit error/state querying but core workflows are well covered.