MCP servers for YNAB workflow automation and Gmail read-only access, with AI-powered Amazon transaction categorization
Server has 7 tools with basic schemas and descriptions, but significant gaps limit production readiness. All tools have descriptions (10-50 chars, mostly short), and input schemas are present with type definitions. However, descriptions lack depth, most are single-sentence statements without context on WHEN to use the tool or what happens on error. Parameter descriptions are minimal (e.g., 'Budget ID or name'). Output schemas are not documented anywhere in the visible code. No error handling guidance, no recovery hints. Tools mix concerns (YNAB + Gmail in one server). Naming is verb-first and clear, but parameters like 'plan_id' and 'tx_type' lack enum constraints despite having fixed valid values. The update_transactions tool accepts an array but provides no per-item success/failure feedback. Overall, definitions are functional but below production baseline.
Get all categories for a specific YNAB budget
Get the full content of a specific Gmail message
Get all messages in a Gmail thread
Get transactions from a YNAB budget, optionally filtered by type (all, unapproved, uncategorized)
List all available YNAB budgets
List Gmail messages matching a query
Update transactions in a YNAB budget with new category assignments
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields. No visibility into what list_budgets, get_transactions, or get_message return.
Parameter 'tx_type' accepts 'all', 'unapproved', 'uncategorized' but is defined as free-form string, not enum. LLMs may hallucinate invalid values like 'pending' or 'draft'.
Descriptions are too short (10-50 chars). Most lack context on WHEN to use the tool, what it modifies, or what to do on error. E.g., 'Get transactions from a YNAB budget...' does not explain whether results are paginated or capped.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 49 | 2025-06-18+ | v2 |
update_transactions accepts array of transactions but provides no per-item success/failure feedback. If 1 of 50 updates fails, LLM cannot determine which ones succeeded.
Server mixes YNAB and Gmail tools in one MCP server. This violates single-responsibility principle and makes the server harder to reason about. Consider splitting into separate servers.