Loading ToolBench
Unofficial local Model Context Protocol server for the monday.com API. Runs on your machine, talks straight to monday.com, keeps your token local.
15 tools with mostly complete schemas and descriptions. Naming follows verb_noun convention well (monday_get_*, monday_list_*, monday_create_*, etc.). Descriptions are present and adequate (avg ~80 chars), but lack LLM-optimized guidance on WHEN to use each tool vs. alternatives. Parameter descriptions are present but sparse, many lack format/constraint details. Output schemas are not documented in the source. Error handling and recovery guidance are absent. No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels. monday_raw_query and monday_ui_action are particularly under-specified.
Create a new item on a board
Post an update/comment on an item
Archive an item
Get board details including groups, columns, and column options
Get full details for specific items by ID
Get the current user's account information
List boards accessible to the current user
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields (e.g., what fields does monday_get_board return? Does it include column IDs needed for monday_create_item?). Breaks tool chaining.
monday_raw_query and monday_ui_action lack sufficient description. 'Execute a raw GraphQL query' and 'Read an item card or post an update' are too vague. LLMs cannot determine when to use raw_query vs. structured tools, or what ui_action actually does with the desktop app.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE). LLMs cannot infer safety properties and may retry destructive operations or assume read-only tools are safe to call repeatedly.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 69 | 2026-07-28+ | v2 |
List items on a board with pagination
List updates/comments on an item
List users on the account
Navigate the monday.com desktop app to a URL
Execute a raw GraphQL query against the monday.com API
Convert monday.com IDs into URLs
Read an item card or post an update via the desktop app UI
Update an existing item
Parameter descriptions lack format/constraint details. 'values' in monday_create_item is described as 'Column values keyed by column ID or title' but does not specify: what types are valid? What happens if a value doesn't match the column type? How are dates, dropdowns, or linked items formatted?
No error handling or recovery guidance. If monday_create_item fails due to invalid column value, the LLM has no guidance on what to do next (call monday_get_board to inspect column types? Retry with a different value?). Errors should include actionable next steps.