FAQ and support

Reading time 4 min

Use the symptom-specific pages first: Build and reader issues, Editor and account issues and Publish and deployment issues. Preserve unsaved work and inspect current state before repeating actions.

Product and installation

Question Answer
Is V2 package 2.0? No; this pin is 1.1.4
Do I need the editor? No; files/Git and build work
Does static hosting run it? No; separate Node writing process
What runtime? Node24+ and available npm
Does this showcase offer editing? No; public documentation only
Is an offline installer bundled? No

The editor is included but optional. To use it on your own installation, create an administrator and start its private server with the correct runtime/users-file settings. Never expose the writer process just because the public reader has a domain. Complete setup separates the installation paths.

Writing and collaboration

Question Answer
Can editors Publish? No; admin only
Can I upload images there? No built-in binary uploader
Is a draft server history? No; local browser recovery
Is Save conflict-free? No general revision check
Can editors write anywhere? Assigned top-level sections only
Does hidden mean private? No; still built/indexed

Add actual static assets through the server/Git workflow and rebuild. Use Git/backups for source versions and deletion recovery. Ordinary Save, metadata revision checks and browser drafts are different mechanisms; none supplies real-time coauthoring. Capabilities and limits gives the full boundary.

Publication and hosting

Question Answer
Does Publish push to Git? No
Does Publish deploy remotely? No; local output switch
Is that switch atomic? No; two renames
Is previous output a full backup? No; runtime/source separate
Does a green build check links? Not every authored link
Can I host at a prefix? Review generated and authored paths

Your hosting/CI/upload process handles the reviewed artifact. DNS, TLS, private reader identity and CDN policy belong to that infrastructure. Check actual remote home/deep/404/search/assets after deployment; a local Publish success is not remote verification.

Markdown, language and network

Question Answer
Are stock fonts/libraries local? Yes, bundled assets
Does that make everything offline? No; remote content/install/fetches differ
Why preview differs? Reader plugins/pipeline differ
Is raw HTML on? No, unless owner file enables it
Are non-Latin URLs guaranteed? No; current ASCII route slug limits
Is this MDX or math support? No built-in dialect promise

Remote images can contact their hosts; favicon refresh is a server fetch; clicked links contact destinations. Local library bundling is not general outbound isolation. Use Markdown lab and inspect actual final reader output rather than enabling raw HTML to repair an unrelated preview mismatch.

Organizational guarantees

No SSO/MFA, clustered writer HA, compliance certification, immutable audit archive or unlimited-scale promise is supplied. External controls can be part of your organizational workflow, but name who owns them and test their actual behavior. Review Validation limits before treating examples or historical measurements as production evidence.

A useful support report

Include Specific evidence
Revision Actual source pin/changed files
Environment Node, OS, browser/device
Reproduction Minimal steps and source
Expected Intended output/response
Actual Exit/status, exact safe message
Location Relative source path/reader route
Scope Synthetic role, not private identity
Visual issue Relevant screenshot and viewport

Inspect existing issues in the actual repository before filing a duplicate when you have access. Verify the repository/destination rather than assuming a generic project name points to its maintainers. Include the smallest synthetic example that reproduces the failure and the actual command used. For publication, add failed stage and reported live state; for search, add query and actual target; for layout, add theme/viewport and longest problematic row.

Keep support evidence private-safe

Never include passwords, temporary passwords, cookies, session/CSRF tokens, API keys, private user files or account-store backups. Remove real usernames, private source, unrelated paths and internal organizational context where unnecessary. Review diagnostic text and screenshot pixels before sharing. Automated redaction cannot certify arbitrary private content is safe.

Replace private values with clearly labeled synthetic data while preserving the behavior. Do not quietly change a path/role so the example no longer reproduces the error. Sharing a minimal error report does not authorize anyone to access your accounts or run production commands.

After a repair

Reproduce the expected result, test the affected allowed/denied boundaries and inspect actual output. State which runtime/browser/environment passed and what remains untested. Keep a focused regression fixture when the engine changes. Close the issue with the actual fix and evidence, not only a successful retry.