Skip to main content
Glama
Brilliant MCP

Brilliant MCP

Official

capture_ui

Screenshots a specified part of Brilliant's UI to an image file, optionally runs commands to set the UI state first, and draws highlight boxes over sub-targets.

Instructions

Screenshot a part of Brilliant's own UI (a toolbar, the palette / color picker, a panel) to an image file. Optionally stage commands to drive the UI into a state first, and draw highlight boxes over sub-targets. target is a capture-registry id such as "full-ui", "top-toolbar", "left-toolbar", "right-toolbar", "bottom-toolbar", "command-palette", or "ai-chat-panel". Chrome only by default on every host (the canvas area is left empty); pass includeCanvas:true (full-ui only) to composite the GPU canvas underneath. Writes lossless WebP by default. Desktop-only: it writes an image file, so it is not available in the web editor.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "target": "<target>",
      "outputPath": "<outputPath>"
    },
    "name": "capture_ui"
  }
}

Input Schema

NameRequiredDescriptionDefault
cropToNoCapture-registry id to crop the result down to. When set, "target" MUST be "full-ui": the full UI is captured (so the panel renders over the real backdrop with true colors), then cropped to this id's on-screen rect. Use this instead of capturing a panel boundary directly, which renders translucent chrome without its backdrop.
formatNoImage format: "webp" (default, lossless) or "png".
targetYesCapture-registry id to screenshot (e.g. "command-palette"). REQUIRED. "full-ui" is the whole window: menus, dropdowns, popovers, tooltips and dialogs included (they mount above the panels), so photograph an open menu with target "full-ui" and cropTo its panel; a panel target never contains its own open menu.
settleMsNoMilliseconds to wait after staging before capturing (default 300).
cropMarginNoLogical pixels to expand the cropTo rect on all sides (clamped to image bounds). Default 0.
highlightsNoBoxes to draw over the capture. Each is either {target: "<registry-id>"} or {rect: [x, y, w, h]} in captured-image logical pixels.
outputPathYesAbsolute file path to write the image to. REQUIRED.
pixelRatioNoDevice-pixel scale for the capture (default 2).
dimBackgroundNoDim everything outside the highlight rects to 35% black. Default false.
includeCanvasNoComposite the GPU canvas front buffer under the chrome. Only meaningful for target "full-ui". Default false.
stageCommandsNoOrdered commands to drive the UI into the desired state before capturing. Same command IDs execute_commands accepts. Executed sequentially; a failure aborts the capture.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.10

TDQS

A4.3/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare readOnlyHint=false, destructiveHint=false and openWorldHint=false. The description adds context the annotations do not: it writes an image file, so it is desktop-only and unavailable in the web editor; lossless WebP is the default format; chrome only by default, with the canvas left empty unless includeCanvas is set. It does not say what happens on an existing outputPath (overwrite) or how failures surface beyond aborting.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

It is front-loaded with purpose, then rules in decreasing priority, and no sentence is filler. It is dense with parenthetical clauses, but each one carries a constraint used to make the call correctly.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

The tool has 11 parameters and no output schema. The description covers the non-obvious interactions (cropTo with full-ui, includeCanvas, settle timing, staged commands aborting the capture), which is what an agent needs before calling it. The remaining gap is what the returned result looks like and how output files are handled on overwrite.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so 3 is the baseline. The description still adds real meaning: concrete registry ids for `target`, the constraint that `cropTo` requires `target: full-ui`, and the `includeCanvas` full-ui-only restriction. Most of its value is convenience for choosing a target rather than overlooked parameter semantics.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The first sentence gives a specific verb (screenshot) and resource (Brilliant's own UI: toolbar, palette/color picker, panel) plus the output (image file). The staging and highlight capabilities are named immediately, so an agent knows this is a capture tool, not a generic render or export tool.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It states the key routing rule explicitly: use `cropTo` with `target: full-ui` rather than capturing a panel boundary directly, because a panel target never contains its own open menu. It also says stageCommands share IDs with execute_commands. It does not state when to prefer a sibling such as render_ui instead, which keeps it short of a 5.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.