Price terms as published

Each line is the vendor’s own wording, split from the price capture checked on 2026-09-24. Nothing here is rounded, converted or re-worded — the full sentences stay in the review below. Lines that carry only a price continue the plan named on the line above, in the vendor’s own order.

2 price terms for Archify
#	Plan / term	Checked
1	Free (MIT)	2026-09-24
2	relies on your own AI agent subscription/API	2026-09-24

One-liner: Archify is an open-source agent skill: install it once, then tell Claude Code, Codex CLI, Cursor, or OpenCode “diagram this” — the agent produces typed JSON IR and Archify compiles it into a self-contained, interactive HTML diagram (architecture, workflow, sequence, data-flow, or lifecycle) with motion, themes, share cards, and deterministic validation checks. One sentence in, an explorable system map out.

Verified facts (GitHub API + raw README + releases, 2026-09-24)

Attribute	Value	Source
GitHub stars	70,713 (4,752 forks, 171 watchers tier per page)	`tt-a1i/archify` via GitHub API, 2026-09-24
Open issues vs PRs	API combined 159 — same-day search-API split: 96 issues + 63 PRs	GitHub API + search API
License	MIT	GitHub API + README license section
Language / stack	JavaScript (Node.js rendering + validation system)	GitHub API + README
Created / last push	2026-04-15 / 2026-09-24	GitHub API
Latest release	v2.16.0, published 2026-08-30; README badge shows dev version v2.17.0-dev.1	GitHub Releases API + README
Install	`npx skills add tt-a1i/archify -g`	README “Quick start”
Agent compatibility	Cursor, Claude Code, Codex CLI, OpenCode (README; more integrations listed under install options)	README
Diagram types	Architecture, workflow, sequence, data-flow, lifecycle	GitHub API description + README
Output	One self-contained HTML file; PNG / SVG / WebM / 1200×630 share-card export	README
Validation	Typed JSON IR, deterministic validation checks, “validation receipts” per scenario, 11 checked-in gallery scenarios	README + Proof Lab
Source evidence	Optional `SRC n` nodes link to Git-verified files and line ranges pinned to one public commit	README
Trending	#1 all-language GitHub Trending repo of the week (creator screenshot, Sep 1, 2026); star-history global rank #294 as of Sep 19	README “Community and recognition” + star-history.com

What it actually is (and what it is not)

Archify is not another drawing app and not a Mermaid renderer. The README describes it as “a Node.js rendering and validation system”: the LLM never hand-draws the picture — it emits typed JSON IR, and Archify compiles that into HTML/SVG deterministically. That split is the whole design: models are bad at pixel layout but decent at structured descriptions, so structure is the model’s job and presentation is the compiler’s. The output opens anywhere (viewing needs no install), supports dark/light themes, guided chapters, route/reach tracing (“highlight the path from web app to database”), and snapshot-vs-snapshot architecture deltas.

The genuinely unusual feature is source evidence: on request, nodes mark themselves SRC n and open Git-verified files and line ranges pinned to one public commit — a diagram you can audit against the repo. The README’s own case study traces mco-org/mco at commit 9f1a1cf into a runtime map. Note the README’s own care on this point: reach share cards “capture that exact reading without claiming runtime impact” — it diagrams authored relationships, not observed behavior.

The numbers behind the hype

The GitHub API count today is 70,713 stars — third-party snapshots show how vertical this was: star-history.com logged 67.1k with global rank #294 on September 19, and a coddykit write-up cites 36,537 stars with +3,722 in a single day when it first trended. Roughly double in five days, with the repo pushed the same morning this page was checked. Sponsors/featured partners in the README include Kimi Work (plugin-store integration), Supercode, and EverMind’s Raven — these are sponsorships, not pricing.

The catches

It’s a skill, not an app. No agent tool, no Archify value — the install path assumes Claude Code / Codex CLI / Cursor / OpenCode and the LLM calls are billed by whichever agent you already pay for. For non-developers, the Kimi Work plugin-store route is the only no-terminal path the README shows.
Validation ≠ truth. The deterministic checks validate structure (typed IR, layout, connections); they cannot tell you the model read your codebase correctly. For decision-grade maps, use the evidence mode or check nodes against the repo yourself.
Release cadence vs README. Latest tagged release is v2.16.0 from August 30 while the README carries a 2.17.0-dev.1 badge and the repo pushes daily — you are effectively tracking main, which is normal for skill-repos but worth knowing if you need pinned versions.

Who should care

Engineering writers / tech YouTubers / docs people: a prompt-to-shareable interactive diagram with export cards is a content pipeline shortcut, and the output HTML can live on any host (it is one file).
Dev-team communicators: the delta view (snapshot A vs snapshot B) and Git-pinned evidence nodes are aimed squarely at “explain what changed in the architecture” reviews.
Everyone else: if you just need a static box diagram occasionally, Mermaid-class tools remain simpler; Archify’s value shows when diagrams are interactive, repeatedly regenerated, or need to be defensible.

Sources

https://api.github.com/repos/tt-a1i/archify — stars, forks, license, created/last-push, topics (2026-09-24)
GitHub search API — open issues (96) / open PRs (63) split, same day
https://github.com/tt-a1i/archify/releases — v2.16.0 (2026-08-30) and the release feed (2026-09-24)
Raw README on main, fetched 2026-09-24 — install command, supported agents, evidence and validation features
star-history.com — 67.1k stars, global rank #294 (Sep 19 snapshot); coddykit.com — first-trend daily gain (secondary)

Remotion: Programmatic React Video — the same “code as the source of truth, renderer as compiler” idea applied to video.
OpenAI Plugins: official Codex plugin examples — where agent skills/plugins themselves are catalogued.

Method note: star/fork/issue counts, license, dates from GitHub REST API on 2026-09-24 (issue/PR split via search API the same day); feature claims from the raw README on main the same day; growth figures from star-history.com (Sep 19 snapshot) and coddykit.com as dated secondary sources. Review type is official — this page was written without installing or running the skill.

Verdict

The fastest-growing way to make a system diagram look designed rather than drawn: your agent writes typed JSON, Archify deterministically compiles it into a single interactive HTML file with validation receipts — and nodes can carry Git-pinned source evidence, which no Mermaid-class tool does. 70.7k stars, pushed on 2026-09-24. Two honest catches: you need an agent tool (Claude Code, Codex CLI, Cursor, OpenCode) already in your workflow — it is a skill, not a standalone app — and diagram accuracy is only as good as the model's reading of your system, so the validation layer checks structure, not truth.