Trello MCP Server for Cursor integration with automation scripts. Provides comprehensive Trello board, list, card, and checklist management capabilities.
This server has 46 tools with highly inconsistent quality. Most tools have adequate names (verb-noun pattern) and basic descriptions (50-150 chars), but critical gaps emerge: (1) Input schemas are completely absent from the provided source code, no JSON Schema definitions visible for any tool. This is a HARD FAILURE under the rubric: 'If a tool has NO input schema at all: its schema score MUST be 0.' (2) Descriptions are present but often generic and lack guidance on when/why to use each tool. Many are under 100 chars and provide minimal context. (3) Parameter descriptions in the source are minimal, field 'description' values like 'The ID of the board' appear repeatedly with no constraint guidance. (4) Output schemas are completely undocumented. No tool shows what fields are returned or what data structures agents can expect. (5) Error handling is not visible in the tool definitions provided, no evidence of error categorization, recovery guidance, or user-fixable vs. fatal distinctions. (6) Security: Credentials (TRELLO_API_KEY, TRELLO_TOKEN) are handled via environment variables (good), but no evidence of rate limiting, permission gates, or audit logging in tool definitions. (7) Composition: Several tools overlap in intent (e.g., create_card vs. create_smart_card_from_commit). The 'smart_card' tools call external services (Git commit analysis, file analysis, code review), these dependencies are not documented in descriptions. (8) Risk annotations exist in the evaluation (READ_ONLY, WRITE, DESTRUCTIVE) but are not visible in the source as MCP tool annotations. (9) Most tools operate on the same resource domain (boards, lists, cards), names are mostly clear (get_board, create_card, delete_list) but lack clarity on preconditions and dependencies. Average naming: 18-25 chars ✓. Average description: 60-100 chars, often too brief.
Add an item to a checklist
Add a label to a card
Add a member to a card
Analyze cards and recommend actions
Analyze developer work patterns
Analyze developer work with detailed comments
Analyze a file for Kotlin implementation
Create a new board
Create a label on a board
CRITICAL: No input schemas visible in source code for ANY of 46 tools. The provided tool definitions show parameter names and descriptions in the summary, but no JSON Schema type definitions, enums, constraints, or format specifications are present in the actual codebase. Under the hard scoring rule 'If a tool has NO input schema at all: its schema score MUST be 0', this is a complete failure for schema validation.
CRITICAL: Output schemas are completely undocumented. No tool description or source code shows what data structure agents will receive. For example, 'get_board' returns unknown fields, agent cannot plan downstream operations without knowing response structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 0 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 36 | - | v1 |
Create a board with initial tasks/cards
Create a new card
Create a new checklist
Create a new list on a board
Create an intelligent card from a git commit
Create intelligent cards from recent git commits
Create multiple tasks in an existing board
Delete a card
Delete a checklist item
Delete a checklist
Delete a list
Get available members for card assignment
Get backlog cards sorted by priority
Get board details
Get all labels on a board
Get all members on a board
Get multiple boards
Get details of a specific card
Get all checklists on a card
Get multiple cards
Get details of a checklist
Get details of a specific list
Get cards from a list with detailed information
Get multiple lists
List all available boards
Move backlog cards to TODO based on priority
Move cards based on priority
Move critical cards to TODO list
Remove a label from a card
Review cards for updates
Review Kotlin implementation
Suggest members for a commit
Update a card
Update card description
Update a checklist item
Update a checklist
Update a list
HIGH: Tool descriptions are generic and often lack guidance on when/why to use each tool versus alternatives. For example, 'get_cards' and 'get_list_cards_with_details' both retrieve cards, the description does not explain the difference or when to use each. Similarly, three separate 'analyze_developer_work' variants exist but descriptions do not clarify distinctions.
HIGH: No evidence of error handling guidance in tool definitions. For destructive tools (delete_card, delete_list, delete_checklist) descriptions do not mention recovery strategies, confirmation patterns, or undo mechanisms. Agents have no guidance on which errors are retryable vs. fatal.
MEDIUM: Parameter descriptions lack constraint information. 'board_id', 'list_id', and 'card_id' parameters are described as 'The ID of X' with no format guidance (UUID? alphanumeric? opaque string?). No minimum/maximum lengths specified. LLMs cannot validate input before submission.
MEDIUM: Overlapping tool intent creates agent confusion. 'create_smart_card_from_commit', 'create_card', and 'create_tasks_in_board' all create cards but descriptions do not explain which to use. External service dependencies (Git, Kotlin analyzer, code reviewer) are never documented, agent cannot predict costs, latency, or failure modes.
MEDIUM: No pagination support documented for list operations. Tools like 'get_boards', 'get_lists', 'get_cards' retrieve potentially large result sets but descriptions do not mention page/limit parameters or result counts. Agent could exhaust context or timeout on large boards.
MEDIUM: Tool annotations (readOnlyHint, destructiveHint, idempotentHint) from the evaluation summary are not visible in source code. Risk tags (READ_ONLY, WRITE, DESTRUCTIVE) exist in the summary but are not implemented as MCP tool annotations in the server definition.
MEDIUM: No idempotency guarantees documented. Create tools do not state whether repeated calls with identical parameters produce duplicate records or return existing results. Critical for agent reliability when retrying on transient failures.
LOW: Naming convention mixed. 'analyze_file_for_kotlin' is Kotlin-specific but other tools use generic domain language. Suggests domain-specific tools were retrofitted into a generic Trello server. Confuses whether this is a general Trello tool or a Kotlin development assistant.