Interactive DBML editor and schema visualizer for Cloudflare Workers with MCP workspace tools
Server has 8 tools with explicit registration and visible schemas. Most tools have descriptions (7/8) and input schemas defined. However, descriptions are inconsistent in depth, some are detailed (schema_apply_patch: 380+ chars), others minimal (canvas_focus: 60 chars). Parameter descriptions are present but sparse; many lack format/constraint guidance. Output schemas are not documented in the source. Error handling is present but recovery guidance is minimal. Naming follows verb_noun convention well (workspace_status, schema_edit, notes_apply_changes). No tool annotations (readOnlyHint/destructiveHint) despite clear risk levels (READ_ONLY vs WRITE). Composition is sound, tools are single-purpose and chainable.
Focus the canvas on specific tables or fit all tables in view
Apply changes to sticky notes in the workspace (create, update, delete)
Get an overview of all sticky notes in the workspace
Applies several exact Schema Source Patch replacements atomically: either all patches apply or none do. Reserved for the atomic multi-replacement case where N exact replacements must succeed together. For a single targeted edit, or for replacing the entire Schema Source, prefer schema_edit. Use schema_source_slice first to capture the exact current text for each patch. Requires Canvas Presence and current Workspace freshness. If any expectedCurrentText is missing, ambiguous, or the edited schema has diagnostics, no source is changed and no Canvas update is broadcast.
Replace the entire schema source with new content. Validates syntax and broadcasts canvas updates on success
Get an overview of the current schema including tables, references, and diagnostics
Output schemas not documented. Tools return structured data but LLMs cannot see field types, required fields, or pagination info. Forces LLMs to guess response structure.
Tool annotations missing. schema_edit and schema_apply_patch are destructive (WRITE risk) but lack destructiveHint. schema_source_slice and schema_overview are read-only but lack readOnlyHint. Agents cannot distinguish safe from risky tools.
Parameter descriptions lack format/constraint guidance. schema_source_slice accepts startLine/endLine but no description states '1-indexed' or range bounds. schema_apply_patch patches array lacks item schema details. LLMs cannot validate inputs before calling.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2025-06-18+ | v2 |
Extract a slice of the schema source code by line range, returning exact text for use in patches
Get the current workspace status including schema source, freshness, and canvas presence
Error handling lacks recovery guidance. No visible error responses or retry logic. If schema_edit fails validation, LLM has no guidance on what to fix or which tool to call next.
canvas_focus description is minimal (60 chars). Does not explain when to use it vs other tools, what 'fit all' means, or whether it requires canvas presence. Ambiguous for LLM selection.