mylesmcook/explainer
v1.6.0MIT AND OFL-1.1
Turn "how does X work?" into an interactive, step-by-step explainer page with an animated diagram, lessons and a message inspector that says what each message does and why.
Changelog
1.6.0 - 2026-11-10
- The right panel explains each message: what it does and why it's sent, above its body and field notes.
- A request links to the message that answers it, and a response links back to its request.
- The lesson summary, facts and links move into a collapsed "About this lesson" section above the message.
- Every message now needs
whatandwhy; the checker reports any that are missing. Older topics must add them before they rebuild. - The writing guide covers how to write
what,whyand field notes, and the checker catches more filler words.
1.5.3 - 2026-11-08
- Add an explicit Claude manifest for claude.ai marketplace synchronization.
1.5.2 - 2026-11-05
- Stronger viewer checks: sessions reuse only the owner's recorded port, the proof is bound to that port, the signing key never appears in links, and secrets are replaced whenever a viewer stops.
1.5.1 - 2026-11-05
- The viewer is private: links carry a token, requests without it are refused, and sessions reuse a running viewer only after it proves it holds the token.
1.5.0 - 2026-11-05
- Pages open without anyone starting a server: building or opening an explainer returns a local link from one shared viewer.
- The viewer is reused by every page and session, stops after 30 idle minutes and with its session, and serves only explainer pages to this machine.
- New tool
explainer_openshows a saved explainer.
1.4.0 - 2026-11-05
- Direct text, no blobs: lesson summaries can be 2 to 4 bullet points, sentences are 20 words max, and leads, summaries, steps, asides, callouts and facts have tighter limits.
- Agents do an editing pass before building, using the writer skill when it is installed.
1.3.2 - 2026-11-05
- Phone playback bar: the Log button no longer spills into the next-lesson button on narrow phones.
- Command-line builds keep the replaced page in
.history, likeexplainer_build.
1.3.1 - 2026-11-05
- Every scrolling panel reserves room for its scrollbar, so text no longer shifts when a panel starts to scroll.
1.3.0 - 2026-11-05
- Spatial stability: controls, lessons, toggles and the diagram never move when something changes, at desktop, laptop and phone sizes.
- The lesson list shows every step of every lesson; any step is clickable.
- At 1024 px, lessons with a Before/After toggle no longer collapse the diagram.
- The checker rejects an actor that moves between a lesson's variants or versions.
1.2.1 - 2026-11-05
topicPathonly reads topics inside the explainers folder; paths or symlinks leading out are refused.
1.2.0 - 2026-11-05
- New tools
explainer_listandexplainer_getto find and revise saved explainers. explainer_checkandexplainer_buildaccepttopicPath(a saved.json) instead of the whole topic.- Rebuilding keeps the replaced page and topic in
~/Explainers/.history/; an identical rebuild changes nothing. - Every tool has a "Use this when" description, a closed input schema checked at runtime, full annotations, an output schema and structured results.
- Lesson list: no double divider above a separated lesson group.
1.1.0 - 2026-11-05
- First marketplace release, from explainer-kit 1.1.0.
- Skill plus bundled stdio MCP server (
explainer_guide,explainer_check,explainer_build). - Checker enforces the topic schema, presets, length limits, actor spacing and plain wording.
- Built pages save their topic JSON beside them for later revision.