This document records evidence for the LobeHub MCP score without treating an external marketplace score as a substitute for product quality.
Target listing:
https://lobehub.com/mcp/juyterman1000-entroly?activeTab=score
A GitHub-hosted audit fetched and decoded the public page on 2026-07-13. The listing reported:
isValidated=false and isClaimed=false;toolsCount=0, promptsCount=0, and resourcesCount=0;python and manual);The 45 points were deployment (15), non-manual installation (12), license detection (8), and README (10). Tools and validation are required and were the two missing required items.
LobeHub’s public implementation assigns 100 points:
| Criterion | Weight | Required | Public observation | Repository evidence | Classification/action |
|---|---|---|---|---|---|
| Claimed listing | 4 | No | False; the page requests its exact GitHub badge before claim verification | Exact LobeHub ownership badge is committed in the primary README | Legitimate ownership proof; public claim check remains external |
| Non-manual deployment | 12 | No | Passing | server.json declares PyPI/uvx and npm/npx install paths |
No product change required |
| Any deployment | 15 | Yes | Passing; two methods | Two canonical package deployment options | No product change required |
| Detected license | 8 | No | Passing but stale as MIT | Current repository and package metadata are Apache-2.0 | External metadata refresh required; points already awarded |
| MCP prompts | 8 | No | Zero | Context-optimization and evidence-verification workflows | Genuine capability added and protocol-tested |
| README | 10 | Yes | Passing but stale | Current primary and package READMEs | External metadata refresh required; points already awarded |
| MCP resources | 8 | No | Zero | Bounded entroly://health and entroly://stats resources |
Genuine read-only capability added and protocol-tested |
| MCP tools | 15 | Yes | Zero despite the existing production tool surface | FastMCP tools for optimization, exact recovery, receipts, memory, and verification | Stale or failed external validation; published-artifact validation required |
| Runtime validation | 20 | Yes | False | Stdio initialize/list/get/read/call protocol tests | LobeHub must re-run validation after publication |
A required criterion missing causes grade F regardless of percentage. If all required criteria pass, 80% or higher is A, 60–79% is B, and lower is F.
The ClawHub v1.0.54 install-metadata correction was completed in PR #124. Its
merge commit has a successful ClawHub Public Listing status for
entroly-openclaw v1.0.54. This LobeHub work remains isolated in PR #125 and
does not rewrite the completed v1.0.54 release.
The added MCP prompts are reusable workflows backed by real Entroly tools. The added resources expose only bounded, read-only operational summaries. They do not expose source content, receipt bodies, filesystem paths, credentials, or unbounded logs.
A clean Python 3.12 validation job installed Entroly from the branch and proved that a fresh stdio server can:
entroly://health and entroly://stats;The repository audit script separately computes local readiness and labels it as local evidence rather than the public score. With protocol validation in the same run, the repository has evidence for 96/100 before the external claim point. This is not the current public LobeHub score.
The work is not complete until all of these are visibly confirmed:
isValidated=true;isClaimed=true; andNo release note or PR may claim an improved LobeHub score before that public confirmation.