Loading ToolBench

Reading the current corpus…

mcp-readiness MCP server · D · ToolBench

mcp-readiness

Lighthouse for MCP servers — grade any public Model Context Protocol server against 10 spec-bound readiness criteria (dual-stack protocol entry, tool quality, safety annotations, anti-ghost content, token efficiency, honest errors) + a Claude/ChatGPT directory pre-flight, then hand off to SaSame's free Capability Control Beta. Legacy Gold Rush v1 commands remain compatibility-only. Zero deps, runs anywhere with npx.

2026-07-28+STREAMABLE_HTTP4 tools

51/100

Definition Quality45

Protocol Readiness62

Supportability53

Spec posture

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

Static source inference · medium confidence · detected: stateful session

Deprecated protocol patterns detected

Summary

This is an MCP server that audits OTHER MCP servers, not a domain-specific tool server. The 4 tools are legacy Gold Rush v1 commands for package management. All tools have descriptions (10-60 chars, below the 50-200 char LLM-optimized baseline). Input schemas are present but minimal: gold_rush_start has 2 params with descriptions; others have 1 param each. No output schemas documented. Tool names use verb_noun pattern (gold_rush_*) but are domain-specific jargon ('gold_rush_package_status', 'gold_rush_agent_run') that lack clarity for general LLM use. Error handling not visible in source. No tool annotations (readOnlyHint/destructiveHint) despite risk labels in metadata. Descriptions are functional but terse, e.g. 'read append-only package state' (28 chars) lacks context on when/why to call. Parameters lack format constraints or examples. This server is primarily an audit/CLI tool; the MCP interface is a secondary compatibility layer.

Tools (4)

gold_rush_agent_runwritesource verified43/100

advance the package one deterministic safe step (no payment)

gold_rush_package_statusread onlysource verified42/100

read append-only package state

gold_rush_reportread onlysource verified45/100

produce the Visibility Report

gold_rush_startread onlysource verified50/100

create or identify a Gold Rush package for an MCP URL

Issues (5)

HIGHtool-description ↗

Descriptions are terse (25-35 chars) and lack LLM-optimized context. 'read append-only package state' does not explain when to call this tool, what it returns, or how it differs from gold_rush_start.

gold_rush_package_statusgold_rush_agent_rungold_rush_report

HIGHresponse-shaper ↗

No output schemas documented. LLMs cannot infer what fields to expect from gold_rush_start, gold_rush_package_status, or gold_rush_report responses.

gold_rush_startgold_rush_package_statusgold_rush_agent_rungold_rush_report

MEDIUMtool ↗

Tool names use domain jargon ('gold_rush_*') that is not self-documenting. An LLM seeing 'gold_rush_agent_run' cannot infer it advances a package state without reading the description.

gold_rush_startgold_rush_package_statusgold_rush_agent_rungold_rush_report

Recommendations

Expand tool descriptions to 50-200 chars. Example: 'gold_rush_package_status' → 'Retrieve the current state of a Gold Rush package, including progress, status, and any pending actions. Call this to check if a package is ready to advance.'
Document output schemas for all tools. Specify what fields gold_rush_start returns (e.g. package_id, status, created_at), what gold_rush_package_status returns (e.g. package_id, state, progress), etc.
Add tool annotations to the MCP schema: set readOnlyHint=true for gold_rush_start, gold_rush_package_status, gold_rush_report; set destructiveHint=true for gold_rush_agent_run (if it modifies state).
Rename tools for clarity: 'gold_rush_start' → 'create_or_identify_package', 'gold_rush_package_status' → 'get_package_status', 'gold_rush_agent_run' → 'advance_package', 'gold_rush_report' → 'generate_visibility_report'.
Add parameter constraints: mcp_url should validate URL format; package_id should specify expected format (e.g. UUID or alphanumeric); goal should be an enum of preset options.

Spec posture evidence

Inferred effective spec: 2026-07-28+.

Relies on Stateful initialize / Mcp-Session-Id (removed; protocol is stateless) - make each request self-contained

Score history

Overall score trend

First recorded score · v2 rubric

51/100

Scored	Grade	Overall	Spec posture	Rubric
2026-09-23	D	51	2026-07-28+	v2