Descriptions exist but are verbose (200+ chars) and include implementation details rather than LLM-optimized guidance. No parameter descriptions, no output schemas, no error handling guidance. Tool names are verb-noun compliant but descriptions read like internal documentation rather than agent-facing specs.

• Add explicit input schemas with typed parameters. Even if tools take no user input, declare 'properties: {}' with 'required: []' and document why no params are needed.
• Document output schemas for both tools. board_new should return {board_id: string, status: string}; board_calibrate should return {offset_x: number, offset_y: number, status: string}.
• Rewrite descriptions as LLM-facing specs: 'Create a new blank Freeform board. Call this once per diagram session. Returns board_id for use in subsequent shape_add calls.' (under 150 chars).
• Add destructiveHint: true to both tools in schema annotations, they modify the Freeform app state and cannot be safely retried without side effects.
• Implement error handling: catch subprocess failures, permission errors, and Freeform unavailability. Return structured errors like {error: 'freeform_not_found', recovery: 'Install Freeform from App Store or ensure it is running'}.