Skip to main content
Glama
Brilliant MCP

Brilliant MCP

Official

init

Start a design session and return the session ID, canvas/repo context, and docs; call it first, then pass that ID to every edit or read call to scope state and cursor.

Instructions

Bootstrap a design session. Returns a unique session ID, session context (canvas ID, repo path), and documentation. Call this first. Pass the returned session ID as the sessionId argument on every subsequent edit / read call: this scopes your refs/design-system state AND gives you your own distinct labelled cursor on the canvas (concurrent agents that echo their IDs never share one cursor). Optionally pass a short agentName (2-3 words) to label that cursor. Pass project (a name or key from list_projects) to bind this session to a project OTHER than the one on screen — including a project that is not open, which the agent reaches headlessly, or the absolute path of a project folder Brilliant has not seen yet (a new Git worktree), which it opens and lists. If you CANNOT read images, pass vision: false: preview requests then return a textual render summary instead of an inline PNG. Then write with edit: an html fence for most tasks (no knowledge loading needed), a bl fence (Blueprint DSL) for advanced features and edits, with its knowledge loaded first via get_knowledge.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "init"
  }
}

Input Schema

NameRequiredDescriptionDefault
visionNoWhether this client can READ inline images (default true). Set false if you cannot see images: preview requests (a `preview()` row) then return a compact TEXTUAL render summary: each element's bounds in parent and world space, layout/overflow checks, and a fill/text one-liner, in place of the inline PNG, so you can verify your write landed without seeing it.
projectNoOptional project to bind this session to: an exact name or key from `list_projects`, or an unambiguous name/handle substring, or the absolute path of a project folder (one holding .bl files, or an empty one), which opens in the background and is added to list_projects. Defaults to the project currently on screen. An ambiguous or unknown value is refused with the candidates listed.
agentNameNoOptional short display name (2-3 words) for this agent's canvas cursor. Clipped to 32 characters.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.10

TDQS

A4.6/5.0
Behavior4/5

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

Annotations only declare readOnly=false, idempotent=false, destructive=false, openWorld=false; the description goes well beyond that by disclosing the session-scoping model, per-agent cursor isolation ('concurrent agents that echo their IDs never share one cursor'), headless project opening, background project listing, and schema-refusal behavior on ambiguous project values. The only gap is that it doesn't say whether a session expires or whether it can be re-initialized safely, which matters for a non-idempotent bootstrap.

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?

Front-loaded with purpose and the 'call this first' directive, and nearly every clause carries operational information. However, it is one dense paragraph that folds the `edit`/`get_knowledge` workflow into a session-init tool, with stacked parentheticals that make it slower to parse than necessary.

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

Completeness5/5

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

No output schema exists, yet the description names the return payload (session ID, canvas ID, repo path, documentation) and covers the non-obvious runtime behaviors an agent needs to call this correctly on first turn. Nothing essential is missing for a zero-required-parameter bootstrap tool.

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 coverage is 100% and the schema descriptions for `vision`, `project`, and `agentName` are already detailed, so most of the description's parameter text reinforces rather than adds. It does add rationale absent from the schema — that `project` reaches projects not open and can take an absolute path to an unseen worktree folder, and why the cursor label exists — which lifts it above the baseline 3.

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?

States a specific verb and resource ('Bootstrap a design session') and immediately enumerates what it returns: a unique session ID, session context (canvas ID, repo path), and documentation. 'Call this first' plus the explicit handoff to `edit`/`read` makes it unmistakable against every sibling.

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

Usage Guidelines5/5

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

Gives ordering ('Call this first'), the downstream contract (pass the returned session ID to every subsequent `edit`/`read` call), and conditional branches: use `project` for a different/headless/unseen project, use `vision: false` when the client cannot read images, and use an ```html fence for most writes versus ```bl after `get_knowledge` for advanced work. Explicit alternatives and when-not conditions are both present.

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