FAQ

What does Meridian do?

Meridian is an open-source project operations toolkit. It takes any open-source project as input and produces a full "promotion kit" in a single AI-assistant session — brand name, multilingual READMEs, VitePress docs site, logo, AI-tool context files, SEO/GEO assets.

What concrete problem does it solve?

"The code is done, but README / i18n / docs site / logo / promotion material all need to be built from scratch" is a task that typically takes maintainers one to two days. Meridian compresses it into a single AI session: say one sentence, pick a style, review the result.

Who is it for?

Any open-source project maintainer, especially:

• Developers who maintain multiple projects solo
• Internal tools that just open-sourced and need promotion material quickly
• One-off users who don't want to learn VitePress / GitHub Actions / i18n conventions

How is this different from create-next-app / cookiecutter / copier?

Those are code scaffolds — they produce runnable code repositories. Meridian is a promotion material generator — the input is an already-runnable project, the output is the README / docs site / logo / SEO assets that live around the project.

It does not touch the project source code, only adds around the perimeter.

How do I get started?

1. Clone Meridian locally
2. Open the Meridian directory in an AI coding assistant (Claude Code / Cursor / Windsurf)
3. Paste this sentence:

Please read my project. The project directory is at [your project path]. Understand the project and set up the operations infrastructure for it.

The AI will explore autonomously → propose a style → wait for your confirmation → execute all tasks. See Quick Start for details.

What's the difference between the 4 visual styles?

Style	Fit
Glow	AI / Agent / generative projects
Minimalist	CLI / libraries / docs-first
Dev-native	shells / SDKs / infra
Enterprise	B2B / platforms / compliance

The AI recommends a default based on your project type; you can accept, swap, or say "random". Each style defines a full visual system: palette, logo, type stack, VitePress theme vars, icon style.

Which languages does it support?

i18n ships four languages by default: Simplified Chinese / English / Japanese / Traditional Chinese. Translations are anchored to i18n/glossary.md as the single source of truth, with scripts/check-i18n-drift.py running drift detection.

Are the generated pages discoverable by search engines and AI answers?

Yes. Meridian's task 12 produces full SEO + GEO assets:

• SEO: robots.txt + sitemap.xml + OG/Twitter Card meta + JSON-LD SoftwareApplication schema
• GEO: llms.txt (per llms.txt standard) + llms-full.txt + structured FAQ page

These make pages indexable by Google/Bing and quotable by ChatGPT / Claude / Perplexity when they answer questions.

How do I resume after an interruption?

During execution, Meridian maintains a checkpoint.md in the target project root. After an interruption, tell the AI:

Please read checkpoint.md and continue the unfinished work.

The AI will skip completed tasks and continue from the next step.

Can I add custom styles or new tasks?

Yes.

• Add a style: create a new directory under templates/styles/ with hero.svg / preview.svg / palette.svg / style.md / vitepress-theme.css, then add a row to the style table in PROMPT.md / QUICK_START.md
• Add a task: create templates/tasks/task-NN-xxx.md, then add a section referencing it in PROMPT.md's task list

Meridian is just "prompt + templates" — all extension points are Markdown files.

Was Meridian's own page really built by Meridian?

Yes. The docs site you're on (lordmos.github.io/meridian), the four-language READMEs, the badges, the logo (hero.svg), the AI-tool context files (CLAUDE.md / AGENTS.md / .cursor/ / .windsurf/), and the SEO assets (OG image, sitemap, llms.txt, this FAQ page) were all generated by Meridian running against itself. This repo is its own demo.