MCP server that collects user's next plan or confirmation through terminal window
The ask_me tool has a functional schema and basic description, but significant quality issues prevent it from being production-ready. The tool name 'ask_me' violates naming conventions (no action verb), the description is minimal (72 chars, below the 194-char average), and the parameter description contains formatting instructions rather than semantic meaning. There is no documented output schema, no error handling guidance, and no information about when/why to use this tool. The tool itself is unusual for an MCP server, it's a human-in-the-loop confirmation tool rather than a capability tool, which doesn't match typical agent patterns.
Ask the user for their next plan or obtain the user's confirmation
Tool name 'ask_me' lacks action verb. Should be 'confirm_plan', 'get_user_confirmation', or 'ask_for_next_steps' to signal intent to LLMs.
Tool description is only 72 characters ('Ask the user for their next plan or obtain the user's confirmation'). This is below the 194-char average and provides no information about when to use it, what it returns, or how it differs from human input.
Parameter 'what_we_did' description contains formatting instructions (TL;DR, bullet points) rather than semantic meaning. Description should explain what information the parameter captures, not instruct LLMs on formatting.
No documented output schema. The tool handler calls handleAskMeTool() but the response structure is not documented. LLMs cannot plan downstream steps or validate returned data.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-21 | F | 32 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 38 | - | v1 |
No error handling or recovery guidance. The tool throws generic 'Unknown tool' errors with no actionable guidance for agents.
Tool is fundamentally a human-in-the-loop confirmation tool, not a capability tool. This is an unusual pattern for an MCP server and conflicts with typical agent tool use cases where the tool should perform actions, not ask for confirmation.
Required parameter 'what_we_did' lacks minimum/maximum length constraints. No guidance on expected string length invites oversized or trivial inputs.