Server has 5 tools with complete input schemas and descriptions, but quality is inconsistent. Tool names follow verb_noun convention (list_, get_, register_, send_, delete_). Descriptions are present but brief (avg 95 chars, below 194-char baseline). Critical gaps: no output schemas documented, no parameter descriptions for most fields, vague 'headers' parameter type (string instead of object), no error handling guidance, and veyra_token exposed as parameter (security anti-pattern). Schemas are structurally valid JSON Schema but lack depth. No tool annotations (readOnlyHint, destructiveHint). Composition is reasonable, each tool has single responsibility.

• Move veyra_token to server-side environment variable or vault. Inject it at runtime without exposing it as a parameter. Update descriptions to say 'Requires Veyra commit mode (configured server-side).'
• Document output schemas for all tools. Example for register_webhook: {type: 'object', properties: {success: {type: 'boolean'}, webhook_id: {type: 'string'}, created_at: {type: 'string', format: 'date-time'}, commit_mode: {type: 'string'}}}
• Change 'headers' parameter type from string to object with additionalProperties: {type: 'string'}. Or if accepting JSON string, add format validation and example: 'JSON object as string, e.g. {"X-Custom-Header": "value"}'
• Add tool annotations to ListToolsResponse: readOnlyHint=true for list_webhooks and get_history; destructiveHint=true for delete_webhook; idempotentHint=true for list_webhooks.
• Enhance error responses with actionable recovery. Example: if Veyra token invalid, return 'Veyra token invalid or expired. Free operations (list_webhooks, get_history) do not require a token. For writes, obtain a fresh token from https://veyra.to/tokens'