<- all tokdocs

Jantt JSON Gantt Engine: Why npm install @jantt/core Fails

By , founder of Mavric and Apso

View on TikTok ->

Jantt is an open-source TypeScript library that turns a JSON plan into an interactive Gantt chart, and the repository is real. The install command a TikTok promotes for it, npm install @jantt/core, fails as of 2026-10-27 because no Jantt package has ever been published to npm. The "zero runtime dependencies" claim is also out of date.

What is Jantt?

A TikTok from @reporadarai scrolls through the project's GitHub README and pitches it as "The Declarative JSON Gantt Chart Engine." The repository is AhmadHassan-BTed/Jantt. The GitHub API describes it as "The AI-native JSON Gantt chart engine. Dependency-free interactive timelines with React, script-tag, CLI, topological DAG solver, and formal JSON schema."

The basic facts check out. The project is MIT licensed, written in TypeScript, created on 2026-08-28, not archived and not a fork. Its last push was 2026-10-26, so someone is working on it. A second copy exists at gustavoml12/Jantt with the same description, but the video points at the original.

The idea is that you, or an AI model, write a plan as JSON and the library handles layout, dependencies and the critical path. The repo ships a real schema at schema/jantt.schema.json, a draft-07 JSON Schema titled "Jantt Plan Specification v1."

The audio in the video says "Jant" with one t at one point. The wordmark and the repository path both spell it Jantt, so search for that.

Why does npm install @jantt/core fail?

The end card tells viewers to run npm install @jantt/core. The README quickstart adds npm install @jantt/react @jantt/core and a script tag pointing at https://unpkg.com/@jantt/standalone.

On 2026-10-27 the npm registry returned HTTP 404 for @jantt/core, @jantt/react, @jantt/standalone and the unscoped jantt. A registry search for "jantt" returned zero results. The unpkg URL returns 404 too, since unpkg reads from the same registry. Running the end card command gives npm error 404 Not Found.

The repository shows why. The "Publish & Release" workflow in .github/workflows/release.yml runs only when a tag matching v*.*.* is pushed. The repo has zero tags, zero releases and zero runs of that workflow, so nothing has ever been sent to npm. The README's npx jantt open command fails for the same reason.

The working path today is to clone the repository and build it from source. The README's contributing section covers that. The video does not mention it.

Does Jantt really have zero runtime dependencies?

The video says "Its core engine has zero runtime dependencies and weighs under 14 kilobytes compressed," and the README carries a Dependencies 0 Runtime badge.

The package manifest at packages/core/package.json declares "fflate": "^0.8.3" under dependencies. The code uses it. packages/core/src/share-coder.ts imports deflateSync and inflateSync from fflate, and src/index.ts re-exports that module, so anyone importing the package pulls fflate in.

The dependency arrived in commit a349c2a9 on 2026-09-05, with the message "fix(share): compress shareable links with deflate-raw and enable dynamic live-update sync." That was 44 days before the video was posted.

There is one fair caveat. The Vite build config sets no external list, so fflate is bundled into the built dist files. A consumer of the built bundle loads one file. The accurate description is that the core includes one third-party compression library. The badge is a hardcoded shields.io image, not a generated count.

Is the "under 14 KB gzipped" figure true?

It cannot be checked from the repository. No package has been published, so there is no tarball to measure. The dist folder is not committed. The CI workflow runs npm ci, npm run build, npm run typecheck and npm test, with no bundle-size check. Nothing in the project fails if the bundle grows past 14 KB.

The README's comparison table lists < 14 KB for Jantt against ~200 KB for DHTMLX Gantt, ~35 KB for Frappe Gantt and ~800 KB for Mermaid.js. None of those four numbers links to a measurement, so treat the whole row as the author's estimate.

Does Jantt use 10x fewer tokens than JSX?

The video says "AI models can generate complete timeline data using 10 times fewer tokens than JSX." The README repeats the 10x ratio six times. It gives no token counts, names no tokenizer and shows no before-and-after comparison. The closest thing is a set of benchmark prompts for Claude 3.5 Sonnet, GPT-4o and Gemini 1.5 Pro, which are prompts without results.

The schema-in, JSON-out approach is implemented. The 10x number has nothing behind it.

A smaller problem sits in the schema itself. Its $id, and every README example, uses https://jantt.dev/schema/v1.json. The domain jantt.dev has no DNS records, so tools that try to fetch the schema from that URL fail. The schema is available at its raw GitHub path.

Are the tests really passing?

The README shows a Tests 339/339 Passing badge. It is a hardcoded image that reads the same no matter what CI does.

CI run 40298259947, from the 2026-10-26 push, failed with 2 failed and 341 passed tests across 22 files. Both failures are in packages/core/tests/reconciler.test.ts. Main-branch CI was green from 2026-09-06 through 2026-09-18 and first failed on 2026-10-22, three days after the video posted. So the pipeline was green when the clip went up, though the test count on the badge was already frozen.

The Node.js >=18.0.0 badge also conflicts with the root package.json, which requires Node 20 or later, and with CI, which tests only Node 20.x and 22.x.

The end card showed 1,118 stars on 2026-10-19. The GitHub API returned 4,620 on 2026-10-27.

Bottom line

Jantt is a real, active MIT project with a working JSON schema. You cannot install it from npm yet, because the release workflow has never run. The zero-dependency, 14 KB and 10x-token claims are either out of date or unmeasured, and the passing-tests badge is static.

Sources

Published October 19, 2026. Based on a TikTok by @reporadarai. Full video notes (markdown)