MCP server that exposes VS Code Output channels (Ruff, Pylance, ESLint, Python, etc.) to AI models via HTTP and stdio transports
Server has 4 tools with complete schemas and descriptions. Naming follows verb_noun convention (list_, read_, clear_, search_). Descriptions are present (avg 120 chars) but lack actionable context for LLM selection. Parameter descriptions exist but are minimal (avg 35 chars). No output schema documentation. Error handling returns plain text rather than structured guidance. Missing parameter constraints (enums, ranges) and recovery hints.
Clear the MCP buffer for a channel (does not affect VS Code UI).
List all VS Code Output channels captured. Returns name, line count, last update.
Read buffered output of a VS Code Output channel (Ruff, Pylance, ESLint, Python, mypy, etc). Use list_output_channels for names.
Search ALL captured channels for lines matching a substring.
No output schema documentation. Tools return plain text or JSON strings without declaring expected fields, forcing LLMs to parse unstructured responses.
Parameter descriptions are minimal (avg 35 chars). 'Channel name (case-insensitive)' lacks context on when to use read_output_channel vs search_output_channels or how to discover valid channel names.
Error responses are plain text strings ('Channel not found') without recovery guidance. Should suggest 'Call list_output_channels() to see available channels' to guide LLM recovery.
No numeric constraints on last_n_lines and last_n_lines_per_channel. LLMs could pass negative, zero, or absurdly large values. Should declare min/max bounds.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2026-07-28+ | v2 |
Tool descriptions lack WHEN-to-use context. 'Read buffered output' does not explain when to call this vs search_output_channels, or what 'buffered' means to the LLM.