FAQ
Common Anydocs questions, with the shortest explanation that helps you unblock quickly.
This page is not a full tutorial. It is a troubleshooting entry point. First decide whether you are looking at a publication-state issue, a reader / preview behavior, or an init / import workflow issue.
Why can't I see a page I just wrote in the reader or preview?
Short answer: the most common reason is that the page is not `published` yet. `draft` and `in_review` stay in Studio for editing, but they do not enter the reader, search indexes, `llms.txt`, or `mcp/` artifacts.
Check these first:
- Has the page status already been switched to `published`?
- Is the page linked from navigation for that language?
- If AI or MCP wrote the content, did it also complete the status change instead of only updating the body?
Why doesn't `preview` open a browser automatically?
Short answer: `preview` starts the local reader server and prints the URL, but it does not open a browser for you.
You should expect:
- a Preview URL in terminal output
- the default entry to point at the first published page in the default language
- a fallback to `/{lang}` if there are no published pages yet
Why can I see a page in Studio but not in preview?
Short answer: that is intentional, not a sync problem. Studio is the authoring surface and shows every workflow state. Preview is the reader surface and serves only `published` content.
Use this distinction:
- Studio is for full source visibility, including `draft` and `in_review`
- Preview is for what readers can actually access publicly
Why are imported pages drafts by default?
Short answer: `convert-import` turns imported content into canonical `draft` pages first, so historical content does not become public without review.
Do these steps right after import:
- review whether the slug fits the target language and URL plan
- clean up or complete the body and metadata where needed
- confirm the navigation location before deciding which pages to publish
What's the difference between `search-index` and `llms.txt` / `mcp/`?
Short answer: they are all public outputs, but they serve different consumers.
- `search-index.<lang>.json` mainly powers site search
- `llms.txt`, `llms-full.txt`, and `mcp/` are better suited for external AI and machine-readable use
- if a document is missing from those outputs, go back to page status first and confirm that it is published
Why do some projects contain `skill.md`, while others contain `AGENTS.md` or `CLAUDE.md`?
Short answer: that depends on whether `init` ran with `--agent`.
- without `--agent`, the generated guide file is `skill.md`
- `--agent codex` generates `AGENTS.md`
- `--agent claude-code` generates `CLAUDE.md` and `.claude/commands/`
If an older example says otherwise, trust the current CLI output.
Why does `init` fail?
Short answer: if the target directory already contains `anydocs.config.json`, the CLI refuses to overwrite the existing project.
The safest fixes are:
- initialize into a different empty directory
- if you meant to work on an existing docs project, open it instead of running `init` again