Loading ToolBench

Reading the current corpus…

mcp-mux MCP server · B · ToolBench

mcp-mux

Centralized MCP Server Management - a gateway and desktop application for managing multiple MCP servers with OAuth support, feature set management, approval workflows, and multi-workspace binding

<=2025-11-25STREAMABLE_HTTP2 tools

75/100

Definition Quality83

Protocol Readiness75

Supportability63

Spec posture

Inferred effective spec <=2025-11-25·current baseline 2026-07-28

Static source inference · medium confidence · detected: Logging, Dynamic Client Registration

Deprecated protocol patterns detected

Summary

McpMux exposes 2 meta-discovery tools with well-structured input schemas, clear descriptions, and proper parameter documentation. Both tools follow verb_noun naming conventions (search_tools, list_all_tools) and include detailed descriptions explaining when to use each. Schemas are complete with proper JSON Schema types and constraints. However, output schemas are not documented in the code visible, parameter descriptions could be more granular, and error handling guidance is absent. The tools are discovery/meta-tools rather than domain tools, which limits composition patterns evaluation.

Tools (2)

mcpmux_list_all_toolsread only50/100

List every tool installed in a Space (default: the caller's resolved Space; pass `space_id` to target another), without the current FeatureSet filter applied. Prefer `mcpmux_search_tools` unless you need the full list — this dump is token-heavy. Returns an array of {server_id, qualified_name, description, available}.

mcpmux_search_toolsread onlysource verified85/100

Search the tools installed in a Space by keyword (default: the caller's resolved Space; pass `space_id` to target another), without the current FeatureSet filter applied. Prefer this over `mcpmux_list_all_tools` when you know roughly what you're looking for — it returns only matches, so it's far cheaper than dumping the whole catalog. `query` is matched case-insensitively against each tool's qualified name, description, and server id. Optional `limit` (default 25, max 100). Returns an array of {server_id, qualified_name, description, available}.

Issues (4)

HIGHtool ↗

Output schema not documented in visible code. Tools return untyped arrays of {server_id, qualified_name, description, available} but no JSON Schema is visible for the return structure.

mcpmux_list_all_toolsmcpmux_search_tools

HIGHrecovery-guide ↗

No error handling guidance in tool descriptions. Neither tool documents what happens on invalid space_id, network failures, or empty results. LLMs have no recovery path.

mcpmux_list_all_toolsmcpmux_search_tools

MEDIUMtool-description ↗

mcpmux_list_all_tools description recommends preferring mcpmux_search_tools but does not explain the specific failure mode or cost/performance tradeoff in quantified terms (e.g., 'returns ~200+ tools, ~5KB response').

mcpmux_list_all_tools

MEDIUMtool-description ↗

Parameter 'space_id' in both tools lacks format guidance. No documentation of how to obtain a valid space_id or what happens if omitted on a non-default workspace.

Recommendations

Document output schema as JSON Schema in code comments or return-type annotations. Example: 'Returns: {type: 'array', items: {type: 'object', properties: {server_id: string, qualified_name: string, description: string, available: boolean}}}'. This enables LLMs to extract IDs for follow-up calls and understand available fields.
Add explicit error handling guidance: 'If space_id is invalid, returns 404 with message listing available space IDs. If no tools match, returns empty array (not an error).'. This tells LLMs when to retry vs. escalate vs. gracefully handle empty results.
Enhance 'space_id' parameter description to: 'Target Space id (obtain from mcpmux_list_spaces). Omit to use caller's default Space. If omitted and no default is configured, returns error with available spaces listed.'
Quantify the token cost tradeoff in mcpmux_list_all_tools: 'Returns all ~N tools in the Space without filter; response is ~Xms and ~YKB. Prefer mcpmux_search_tools (returns ~3-20 matches, ~1-5KB, ~50ms) unless you need the complete catalog.' This guides the LLM to pick the right tool.
Document the 'query' matching behavior more precisely in mcpmux_search_tools: 'Keyword is matched case-insensitively. Multiple keywords are AND'd together. Matches against: tool.server_id, tool.qualified_name, tool.description. Returns matching tools sorted by relevance (server_id first, then name, then description).'
Add idempotent or readOnly annotations to both tools (they are GET operations). Example: '{"idempotentHint": true, "readOnlyHint": true}' if tool annotations are supported.

Spec posture evidence

Inferred effective spec: <=2025-11-25.

Relies on Logging (deprecated) - log to stderr or use OpenTelemetry

Score history

Overall score trend

↑ 35 points across a rubric change (v1 → v2)

75/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	B	75	<=2025-11-25	v2
2026-03-09	F	40	-	v1