mdstudio.app

What It Is

Markdown Studio is an AI Markdown editor built for prompt engineers and technical writers. It brings Markdown writing, live preview, token counting, prompt templates, multi-model comparison, and GitHub sync into a single workspace, positioned around “writing prompts, tracking tokens, and shipping faster.” The page emphasizes zero knowledge, no account required, and data processed only on the user’s device.

Core Capabilities

Its AI focus is prompt testing rather than an in-house model. The tool supports OpenAI, Claude, and Gemini, and can run the same prompt in parallel via Battle Mode to compare outputs, tokens, cost, and performance. It also offers autocomplete for 23 prompt tags, 24 templates, variable placeholders, smart copy formatting, and guidance for parameters such as temperature, max tokens, top-p, seed, and model version locking. The editor itself is also fairly complete, with split-screen preview, 40+ slash commands, tabs, Mermaid, LaTeX, presentation mode, and export to PDF/HTML/JSON/Markdown.

Pricing and Privacy

The page explicitly calls it a Free AI Markdown Editor, with no sign-up required and offline availability after the first load. However, it does not clarify whether AI calls include any free quota, nor does it disclose how external model API costs are billed. Privacy is its biggest selling point: rendering, token counting, exports, and LLM calls are all claimed to run locally, while API requests go directly from the browser to model providers without the platform sitting in the middle. There is no user database, and cookie-free Umami is used only to collect feature usage statistics.

Pros, Cons, and Who It’s For

Its strengths are a low barrier to entry and a tight integration of Markdown and prompt workflows, making it especially suitable for multi-model benchmarking, cost estimation, technical documentation, and prompt template reuse. GitHub sync, conflict detection, encrypted tokens, audit logs, PII detection, and snapshot verification also give it some value for team governance. Limitations include the lack of disclosure around the company entity, support channels, enterprise pricing, model key configuration, and Chinese-language support. Output quality still fundamentally depends on the connected models, and deterministic execution may also be affected by model version changes.

Access from China

The crawled text does not provide information on accessibility from mainland China, payment methods, or ICP filing status, so this is considered unknown. Even if the website is accessible, the network and account availability of model services such as OpenAI, Claude, and Gemini in mainland China may become a practical bottleneck. Alternatives include Typora, Obsidian, and VS Code Markdown plugins combined with model platforms, or prompt debugging tools such as PromptLayer, LangSmith, and OpenRouter Playground.