Lint fix or slide syntax?
Repeated slide headings, page separators, notes and interaction fences need studio-aware review. Apply generic lint fixes only to a copy and inspect the resulting deck.
Elysia Tools
Navigation
Workflow Playbook
Review one interactive Markdown lesson, reconcile content and settings, test the actual offline player and deliver audience files separately from private speaker sources.
New to this workflow? Follow the steps in order, starting with the first tool below. Already know what you need? Jump straight to the matching step — every tool in this workflow is listed in the comparison table.
First tool: Markdown Lint and Style Checker
Hubs
| Tool | What it does | Category |
|---|---|---|
| Markdown Lint and Style Checker | Lint Markdown against ~15 markdownlint rules plus 2 Chinese-typography rules, with line-located reports and one-click auto-fix of safe issues | Text Processing |
| Markdown Link Checker | Statically check Markdown links for format errors, broken anchors, and undefined references (no network requests) | Development |
| Slide studio | Turn Markdown into interactive presentations. Edit locally, present offline, or publish an unlisted snapshot. | Document Tools |
| Text Diff | Compare two text inputs and show differences | Text Processing |
Hubs
| Compare two JSON payloads and render a visual HTML diff with path-level additions, removals, changes, and statistics |
| Development |
| Color Contrast Checker | Check color contrast ratio and WCAG compliance for accessibility | Design |
| Accessibility Checker | Detect common WCAG 2.1 accessibility issues in HTML, fetched pages, or design images and return fix-ready guidance | Validation |
| File Hash Verifier | Calculate MD5/SHA1/SHA256 hash values for files and verify against expected hashes | Utilities |
This workflow is for trainers, teachers and technical speakers preparing one interactive lesson, not generating a generic article or replacing a learning platform. Record the learning outcome, approved claims, presenter and expected duration. Identify the audience package separately from the private working source. Remove learner identities, credentials and confidential examples before using helpers that may process inputs on servers.
Keep baseline Markdown and JSON. The studio uses headings, page separators, speaker-note markers and reveal/quiz directives. A general Markdown linter can flag intentional repeated headings or change significant spacing. Work on a copy, explain findings and verify accepted edits in the studio rather than auto-fixing the whole deck blindly.
The Markdown link checker is static: syntax, reference and anchor findings do not establish HTTP availability. Open needed links separately and verify anchors in the actual deck. Prepare a fallback for live destinations; a link in an offline lesson does not make its target available offline.
Use text diff for wording and JSON diff for slide order, layout, theme, transitions and interaction changes. Compare stable revisions, not unrelated exports, and explain regenerated identifiers rather than treating every difference as content. Check quiz choices and correct answers manually against the learning goal; a well-formed quiz can still teach an incorrect fact.
Measure actual foreground/background pairs in the selected theme. Inspect sanitized rendered HTML for common labels and structure problems where applicable. Static tools do not execute reveal sequences, certify focus behavior or guarantee accessibility. Rehearse keyboard access, visible focus, result feedback, contrast on interaction states and readability on a small screen; record unresolved issues instead of claiming certification.
Export HTML and open that exact file in the target browser. Disconnect networking and navigate every slide, reveal and quiz. Test long text, code blocks, navigation boundaries and required media. Record browser, revision and missing resources; embedded player code does not make every linked destination or external asset offline. Resolve required failures or clearly reduce the delivery promise before approval.
Audience-player code strips speaker notes, but source exports can retain them. Inspect HTML and the actual audience view, then inspect Markdown/JSON separately. Never send a notes-bearing source to an audience simply because its preview hid the notes. The studio limit is 80 slides and 2 MB; unsupported inputs need deliberate reduction, not silent omission.
Hash the final files after the last approved edit. Keep an access-controlled presenter package with editable sources and a separate audience package with reviewed HTML. Attach revision, change ledger, rehearsal evidence, known limitations and reviewer. Hashes identify bytes, not teaching accuracy.
Publication is an optional additional step: login to publish a snapshot, then check the exact resulting audience URL and note-free content. Unlisted means not openly listed, not confidential access. Editing the source is not proof the existing snapshot changed; verify the published revision explicitly. Use general Markdown publishing, technical localization or web accessibility workflows for their wider tasks. None replaces rehearsal of this exact interactive lesson.
Workflow playbook
Lint a copy and statically check Markdown links; manually preserve slide-specific syntax and check live destinations separately.
Use the studio to prepare the lesson, compare source text and JSON settings, and explain slide-order, theme or interaction changes.
Check actual theme color pairs and sanitized rendered HTML for common accessibility issues; reserve interaction acceptance for rehearsal.
Export HTML and open that exact file offline; manually test controls, interactions, small screens and speaker-note separation.
Repeated slide headings, page separators, notes and interaction fences need studio-aware review. Apply generic lint fixes only to a copy and inspect the resulting deck.
Use exported HTML for the offline rehearsal. Publishing requires login, but an unlisted audience link is not access control; publish only shareable material.
Contrast and HTML checks locate issues. Keyboard navigation, reveal order, quizzes, mobile layout and offline behavior need actual browser rehearsal.
Hash final files and separate private editable sources from audience delivery and optional publication records.