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

Annotations already provide readOnlyHint=true, idempotentHint=true, and destructiveHint=false, so the agent knows it's a safe, non-mutating call. The description adds valuable context with 'No LLM call' (clarifying it's deterministic and fast) and describes the response structure in some detail (the 'evidence' packet and detail-level options). However, it doesn't go deep into return format specifics beyond the brief list; given annotations cover the safety profile well, the added context is moderate but not exhaustive.

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

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

The description is compact but dense with information. It front-loads the key action ('Orchestrator - start here') and then lists the main outputs in a single breath. It avoids redundancy, uses clear terminology, and every sentence adds value – from the deliverable list to the skip condition to the 'No LLM call' note. It's well-structured for an agent to parse quickly.

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

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

Given the tool's complexity (9 parameters, 7 enums, rich output list) and no output schema, the description covers the essential behavioral contract well: what it returns, how to skip the pick step, and the detail-level modes. It also mentions the 'evidence' packet and reference component sourcing. However, it doesn't explain how parameters interact (e.g., how vibe and color combine) or the exact semantics of 'paper band / display class / accent hue' – but that might be over-specification. It's complete enough for most use cases.

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

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

Schema description coverage is 100%, so each of the 9 parameters is already described in the schema. The description adds context about the overall response and how detail affects the output ('concise' vs 'full'), but it doesn't add meaning to individual parameters beyond that. For instance, the 'vibe' and 'color' parameters are self-explanatory from their enums and descriptions. The baseline 3 is appropriate because the description adds some cross-parameter context but the schema does most of the work.

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

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

The description starts with 'Orchestrator - start here for a brief' which clearly identifies this as the entry point and its primary purpose. It enumerates the specific deliverables (macrostructure pick, exemplars, reference components, palette suggestion, evidence packet), making it distinct from sibling tools like find_components or get_design_system. The verb 'recommend' combined with 'start here' leaves no ambiguity about its role.

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

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

Explicitly says 'start here for a brief' and the 'One call returns...' enumerates what the agent gets. It also states 'Pass a macrostructure to skip the pick step' – a clear condition for when to simplify the call. While it doesn't explicitly name alternatives, the instruction 'source via get_reference_jsx' points to a sibling for a follow-up step, and the overall description makes it obvious this is the first call in a workflow compared to sibling tools like compare or find_similar.

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