Loading ToolBench

Reading the current corpus…

plugin-mcp MCP server · F · ToolBench

plugin-mcp

A Limbo chat plugin that acts as an MCP client, connecting to remote MCP servers via HTTP and exposing their tools as Limbo tools

2026-07-28+STREAMABLE_HTTP0 tools

0/100

Definition Quality0

Protocol Readiness0

SupportabilityNot measured-

Spec posture

Inferred effective spec 2026-07-28+·current baseline 2026-07-28

Static source inference · medium confidence · evidence: Streamable HTTP

Current-spec patterns detected

Summary

Scoring was not performed

Issues (7)

CRITICALtool ↗

Tool 'reload-tools' has no input schema.

reload-tools

CRITICALtool-description ↗

Tool description 'Reload tools' is only 12 characters. This description is trivial and does not explain WHAT the tool does, WHEN to use it, or any side effects.

reload-tools

HIGHtool ↗

Tool name 'reload-tools' is compound (contains 'and' logic implicitly: 'unregister all tools AND re-register them'). Rubric pattern:tool states compound names signal multiple responsibilities and should be split. Additionally, this is a meta-management tool, not a primary domain action.

reload-tools

MEDIUMtool-description ↗

No parameter descriptions for 'reload-tools'. The rubric requires every parameter to have a description explaining what it controls. While this tool takes no parameters, that should be explicitly documented.

reload-tools

HIGHtool ↗

Tool does not document return type or output schema. The rubric (pattern:tool, mxe:response-field-naming) requires that tool responses be documented so LLMs know what to expect and can plan downstream calls. The reload-tools command appears to return an empty string ("") but this is not documented.

reload-tools

HIGHrecovery-guide ↗

No error handling or recovery guidance in 'reload-tools'. If registration of any MCP server fails, the tool logs the error but does not guide the LLM on how to recover. Promise.allSettled suppresses errors silently.

reload-tools

HIGHconfirmation-request ↗

Tool performs a destructive operation (unregisters all tools) without confirmation or dry-run support. Rubric pattern:confirmation-request states irreversible operations should support a dry-run or confirmation step to prevent catastrophic errors.

reload-tools

Recommendations

Expand the tool description from 'Reload tools' to at least 50-100 characters explaining its purpose, side effects, and when to call it. Example: 'Reload all MCP tools from configured server URLs. Unregisters current tools and re-initializes connections. Use after updating server configurations.'
Define an explicit input schema for reload-tools (even if empty). Use JSON Schema format: { type: 'object', properties: {}, required: [] }. Make it clear the tool takes no parameters.
Document the return type and output structure. Currently the tool returns an empty string, define whether it should return success/failure status, list of reloaded tools, or error details.
Add error handling that returns actionable messages. Instead of Promise.allSettled silently suppressing errors, collect failures and return a message like: 'Reloaded tools. Errors: Server X failed (reason). Manually check server configurations.'
Add a 'dry_run' boolean parameter to preview what would be reloaded without actually unregistering tools. Alternatively, require explicit confirmation before unregistering all tools.
Document the destructive side effect (unregisters all tools) prominently in the description. Mark with a [DESTRUCTIVE] tag to alert agents and humans.
Consider renaming to 'sync_mcp_tools' or 'refresh_mcp_tools' if the intent is to synchronize with configured servers rather than a full reload. 'reload' implies a restart, which may not match user expectations.

Score history

Overall score trend

↓ 19 points across a rubric change (v1 → v2)

0/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-22	F	0	2026-07-28+	v2
2026-03-09	F	19	-	v1