Archify
tt-a1i/archifyA skill that turns a description or a codebase into a validated, interactive standalone HTML diagram or one-page explainer.
Turns complex answers into a one-page HTML explainer: the model writes a short Markdown draft, and a bundled CLI generates layout, diagrams and dark mode.
A focused, benchmarked skill that measurably cuts tokens and latency versus hand-written HTML, with a sensible proactive-use rule and clean MIT licensing.
The author changed the skill files after our review (noticed Oct 29, 2026). The review describes the earlier version; we recheck changed skills weekly, the most significant changes first.
The idea is to stop the model from hand-writing HTML, CSS and SVG, which the README's own token count found makes up about 79% of a hand-written explainer page. Instead the model writes a short extended-Markdown draft, and the bundled am CLI (Node.js, no separate npm install) turns it into a complete offline HTML page with layout, colors, dark mode and diagram coordinates computed automatically. The skill's own benchmark (in bench/) reports about 8 times fewer output tokens and 2.6 times faster generation than asking the model for HTML directly, for the same questions, with the raw token counts and corpus available to reproduce.
The skill is meant to trigger proactively: whenever an answer has three or more interrelated concepts, a flow or protocol, a multi-dimensional comparison, a hierarchy, or when the user says something like 'draw it' or 'I don't get it', not only when explicitly asked for a page. It explicitly tells the agent not to hand-write HTML, CSS or SVG itself. It also supports an always-on mode, explainer videos (described as saving even more tokens in the README's own test), and config/clean/update subcommands handled by a separate reference file. One script ships inside the skill folder. The README lists Claude Code, Codex, Cursor, OpenCode, Pi and DeepSeek Harness as working hosts, and bundles the Bangers font under a separate open font license for one theme.
One skill at about 25 KB with 2 reference files and one bundled script (the rendering CLI itself lives outside the skill-metrics count); no test suite, though the repo ships a reproducible benchmark corpus.
Counted from the repository file tree. How we score depth.
A skill that has the agent write a short Markdown draft and a bundled Node CLI turn it into an HTML explainer page or video; its description tells the agent to use it proactively and liberally on explanatory answers. The CLI itself was not available to us. It writes under ~/.answer-me-with-html, opens pages in your browser by default, checks GitHub once a day for updates (opt-out available), and can send video narration to ElevenLabs if you set a key.
am config set update_check false or AM_NO_UPDATE_CHECK); am theme add downloads a theme from answer-me-with-html.com.skills/answer-me-with-html/references/settings.md:12npx skills add QingYunA/answer-me-with-html and npx skills update answer-me-with-html -y (unpinned installer), and the agent-assisted install may install Node.js 20+ with brew or winget.README.md:107The repository (364 files) on Oct 29, 2026. Not a full review: 349 of 364 files were not downloaded (large repository; we reviewed the skill folders); it relies on code outside the repository: scripts/am.mjs (the bundled CLI) and the rest of the repo: 349 readable files were not available to us (including INSTALL.md, plugin manifests and workflows); only README.md, SKILL.md and two reference files were reviewed; npm package skills (vercel-labs/skills installer); Remote theme catalog at answer-me-with-html.com and ElevenLabs.
NVIDIA SkillSpector static scan: 11 pattern hits (MCP Rug Pull 4, Rogue Agent 3, Data Exfiltration 3, Anti-Refusal 1). 'update this skill' is the documented update flow (the skill says to ask before running updates or clean-up); 'no warnings' is formatting guidance; the star-history hit is an image badge in the README; 'write to ~/.answer-me-with-html' is the page output folder.
A review, not a certification. Skills change after we check them — read the files before installing.
MIT — commercial use allowed; one bundled theme font (Bangers) is under the SIL Open Font License 1.1.
Not really. It renders one answer page per turn in an interactive conversation; it has no loop or long-running task.
A skill that turns a description or a codebase into a validated, interactive standalone HTML diagram or one-page explainer.
Agent skills from AntV's infographic engine: one writes a self-contained infographic from your text; four others help contributors extend its templates.
A diagram skill that writes 45 diagram types as one self-contained HTML and SVG file in your brand, with onboarding, import and PNG export.
It is an agent skill that turns a complex answer into a single, readable, offline HTML page instead of a wall of plain text. The model writes a short Markdown draft, and a bundled CLI builds the page's layout, diagrams and styling.
You need Node.js 20 or newer; there is no separate npm install step since the CLI is bundled in the skill. The README's recommended path lets your agent install it for you after you ask it to.
The author's own benchmark, reproducible with node bench/corpus.mjs, found about 8 times fewer output tokens and roughly 2.6 times faster generation across 3 topics and 3 runs, because the model writes only Markdown instead of hand-rolled SVG coordinates and CSS.
It is meant to trigger proactively whenever a page would help more than plain text (multi-step flows, comparisons, hierarchies, diagnoses), not only when explicitly requested, but skips trivial answers or small talk.