What did Hlido score claude-mermaid?

claude-mermaid scored 80/100 (STEADY) in Hlido's independent, hands-on review.

How much does claude-mermaid cost?

Hlido recorded the pricing model as: Open source. Derived from evidence Hlido holds, not vendor-supplied; verify current prices on the vendor's pricing page.

Does any vendor pay Hlido for placement?

No. Hlido takes no money from the agents it rates — scoring weights stay private and the evidence behind every verdict is public.

claude-mermaid

Name: claude-mermaid review
Item: claude-mermaid
Rating: 80
Author: Hlido Editor

Coding · tested 2026-09-01 · re-test due 2026-12-01 · by the Hlido desk, not the vendor

In short: A two-tool MCP server that does one narrow job — live Mermaid previews inside Claude Code — and documents it better than most products ten times its size.

5 PASS · 0 FAIL of 5 public-surface claims

Visit claude-mermaid → scorecard.json

Quick answer

claude-mermaid scores 80/100 (STEADY) on Hlido’s independent, hands-on test (reviewed 2026-09-01). STEADY (80) because the public surface does the hard part well: the full two-tool contract is published with parameters, types and defaults, installation is documented for three distinct client paths, the licence and aut Pricing: Open source (free entry point documented).

Hlido Editor · 2026-09-01

claude-mermaid solves a specific irritation in agent-assisted work: you ask a coding agent for a diagram, it writes Mermaid source, and you then leave the loop to look at it. This server closes that loop. It renders the diagram with @mermaid-js/mermaid-cli, serves it from a local server on port 3737, and pushes a WebSocket reload on every subsequent edit, so the browser tab keeps pace with the agent. The public surface is unusually strong for a single-maintainer open-source tool. It publishes the complete tool contract — both tools, every parameter, type and default, in a table on the landing page — which is exactly the artifact another agent needs to decide whether to call it, and it is the artifact most MCP servers omit. Installation is documented for three client paths (Claude Code plugin, npm, generic MCP clients), the licence is stated (MIT), the author is named, and a standalone gallery mode is offered for use without an agent at all. The limits are structural rather than sloppy. Everything is local: there is no hosted or remote surface, so the server is only as available as the developer's own machine, and it inherits a Node plus mermaid-cli dependency chain the page does not version. Live reload is scoped to SVG; PNG and PDF are render-on-save. And the surface carries no adoption, release-cadence or reliability evidence at all — no changelog, no version history, no downloads figure, no third-party usage — so nothing here speaks to whether the project is maintained beyond the day it was read. That is the gap between a well-documented tool and a dependable one, and it is the reason this sits at STEADY rather than higher.

Why STEADY

STEADY (80) because the public surface does the hard part well: the full two-tool contract is published with parameters, types and defaults, installation is documented for three distinct client paths, the licence and author are stated, and the product's single claim — live reload of a rendered diagram as the agent edits it — is explained mechanically enough to be checked. Not VITAL because nothing on the surface evidences durability: no changelog, no release history, no version or download signal, no third-party usage, and a hard dependency on a local Node plus mermaid-cli toolchain that is never pinned. High documentation quality is not the same as demonstrated reliability, and only the first is visible here.

Public-surface checklist

PASS Homepage loads (required)
PASS Primary value prop (required) — 'Ask Claude for a diagram. Watch it render as you refine it.'
PASS Cta present (required) — 'Install it' / '/plugin marketplace add veelenga/claude-mermaid'
PASS Pricing or access — 'Free and open source · MIT license · works with any MCP client.'
PASS Evidence or demo — Worked example of an OAuth sequence diagram rendered to localhost:3737/oauth, plus a full two-tool parameter reference

What we saw

4 screenshots captured by the Hlido engine during the reviewed run (run-0884a400e2e8d8fc-veelenga-github-io). Our own captures — not vendor marketing material.

claude-mermaid — run screenshot 1 (home.png) — home.png

claude-mermaid — run screenshot 2 (page_main.png) — page_main.png

claude-mermaid — run screenshot 3 (page_top.png) — page_top.png

claude-mermaid — run screenshot 4 (page_how.png) — page_how.png

What it does well

Publishes the complete tool contract on the landing page — both tools, every parameter with type and default
Explains the render loop mechanically (tool call, mermaid-cli render, local serve on :3737, WebSocket reload) rather than asserting it
Documents three separate install paths: Claude Code plugin, npm, and generic MCP clients
States licence (MIT), author, and links both GitHub and npm from the landing page
Supports concurrent previews via named preview_id, so several diagrams can be worked in parallel
Offers a standalone gallery mode (claude-mermaid --serve) that works with no agent attached

What it fails at

No adoption or reliability evidence on the surface — no changelog, release history, version list, or download figure
Entirely local: no hosted or remote endpoint, so availability is the developer's own machine
Unpinned dependency chain — requires Node and @mermaid-js/mermaid-cli, neither version-constrained on the page
Live reload is SVG-only; PNG and PDF render on save rather than continuously
Fixed default port 3737 with no documented conflict or reconfiguration story
Single-maintainer project with no stated support or maintenance commitment

Best for

Claude Code users who iterate on architecture, sequence or ER diagrams during a coding session
Developers who want diagrams-as-code reviewed visually without leaving the agent loop
Teams already standardised on Mermaid who want a local, no-account rendering path
Anyone needing SVG/PNG/PDF export of agent-authored diagrams into a repo path

Not recommended for

Hosted or CI environments — the server assumes a local machine and a local browser
Diagram formats other than Mermaid (no PlantUML, Graphviz or D2 path)
Teams requiring a supported dependency with a published maintenance or security commitment
Shared or multi-user rendering — previews are per-machine, on a fixed local port

Pricing & access

ModelOpen source
Free entry pointYes — a free tier or open-source edition is documented
Pricing findable on the public surfacePASS 'Free and open source · MIT license · works with any MCP client.' (tested 2026-09-01)

Derived from Hlido-held evidence only (engine checklist + editorial text); quotes are verbatim from the scorecard; not vendor-supplied; re-derived daily. Verify current prices on the vendor's pricing page. Last verified 2026-09-01.

Compared to

Excel MCP Server visual-feedback-loop
Both are narrow, well-documented single-purpose MCP servers with published tool contracts. The Excel server acts on a durable artifact the user keeps; claude-mermaid renders an ephemeral preview the user looks at. Choose by whether you need the agent to change a file or to show you something.
MCP-NixOS iteration-over-lookup
mcp-nixos exposes a large external knowledge surface for the agent to query; claude-mermaid exposes two tools and a rendering loop. mcp-nixos is the better model for read-heavy lookup, claude-mermaid for tight human-in-the-loop iteration.

Agent relevance

CLI MCP Behavioral-testable

Agentic-Commerce Readiness 66/100 · INTEGRABLE

Independent readiness for agent delegation & transaction. How it’s scored · check live

Native MCP server. Install as a Claude Code plugin or run the claude-mermaid command from any MCP-compatible client; the agent then calls mermaid_preview(diagram, preview_id, ...) to render and mermaid_save(save_path, preview_id, ...) to write the result to disk. Both tools and all their parameters are documented publicly, so an agent can bind to the contract without reading source. Local-only — the agent and the server must share a machine.

Agent-friendly score: 9/10

Evidence

Exposes exactly two MCP tools, mermaid_preview and mermaid_save, with full parameter tables — source (2026-09-01) verified
Live reload pushed over WebSocket on every agent edit; local server at localhost:3737 — source (2026-09-01) verified
MIT licensed, free and open source, built by Vitalii Elenhaupt — source (2026-09-01) verified
Installable as a Claude Code plugin, via npm, or with any MCP-compatible client — source (2026-09-01) verified
Exports to SVG, PNG and PDF; live preview is SVG-only — source (2026-09-01) verified
No changelog, release history or adoption figure published on the landing page — source (2026-09-01) verified

scorecard.json · transparency passport · registry · methodology

More: compare agents · best of · developer tools · incident registry

Verdict by Hlido Editor, our automated editorial system · Method: public-surface-tier-1+editorial-narrative-v2 · Methodology version 2026.05 · Next review due 2026-12-01

How this page was produced. The scores, claim verdicts and evidence come from automated hands-on testing of the product’s public surface. The written analysis is drafted by an AI system, and pages publish without a person reviewing each one. Hlido publishes this record and answers for it — tell us if anything here is wrong and we will correct it.

Embed this trust badge

Live, always-current independent score — free to embed on your site or README. No vendor pays for placement.

Markdown

[![Hlido trust score](https://hlido.eu/badge/veelenga-claude-mermaid.svg)](https://hlido.eu/check/?agent=veelenga-claude-mermaid)

HTML

<a href="https://hlido.eu/check/?agent=veelenga-claude-mermaid"><img src="https://hlido.eu/badge/veelenga-claude-mermaid.svg" alt="Hlido trust score"></a>