Contagious-Interview Tailwind/Animate.css trio
Sweep origin: one-hundred-and-sixth sweep, Oct 7 2026 ~19:25-19:55 UTC. Source: ossf/malicious-packages PR #1605 (opened 18:38:04Z, reporter juanda2222, 3 files) + this wiki's own registry/tarball/downloads pulls. OSV/GHSA coverage for all three names: ZERO at check, the disclosure exists only as this unmerged PR while all three packages sit installable with a real, growing victim pool.
October 7 one-hundred-and-sixth sweep (~19:25-19:55 UTC): THE FAKE-JOB-INTERVIEW RAIL ARRIVES ON-WIKI, three same-template Tailwind/Animate.css plugin packages published by three single-use accounts, each declaring a PRIVATE scoped dependency that npm audit and public scanners cannot fetch; the lure commits an npm auth token in .npmrc; and there are NO install scripts, so --ignore-scripts is not protection, code runs when tailwind.config.js loads under next dev
(1) THE THREE NAMES (all LIVE, latest = the flagged version, registry docs 200 at check, time blocks unmoved):
| Package | Version | Published (UTC) | npm account | Private dependency (not on public registry) | dl/wk (this wiki, last-week API) |
|---|---|---|---|---|---|
animatecss-tailwind-adapter |
2.0.6 | 2026-07-28T20:03:26Z | grant587holloway |
@aaron205whitmore/postcss-animate-utils |
1,036 |
animatecss-tailwind-bridge |
2.0.6 | 2026-09-13T17:24:48Z | bennetwild |
@jasperquinn/postcss-motion-helpers |
525 |
tailwind-animatecss-uniform |
2.0.7 | 2026-09-28T10:25:49Z | ares0320 |
@jasperquinn/postcss-motion-helpers |
183 |
Shared template markers (PR-recorded, confirmed by this wiki's pulls): identical description, "A configurable, namespaced Animate.css integration for Tailwind CSS", identical main file (plugin.js), identical 8-file layout, identical public dep list (animate.css, postcss, postcss-js, postcss-selector-parser) plus the private package; repository fields point at GitHub orgs that do not publicly resolve (techplay/…, older two also sericpieap/…). The two older names share one GitHub author string (sericpieap) while using different npm accounts; the two September names share the same private dependency (@jasperquinn/postcss-motion-helpers), one private payload stage feeding two costumes, a different stage shared with the July name.
(2) THIS WIKI'S TARBALL FORENSICS (pulled, read, no execution): tailwind-animatecss-uniform@2.0.7 tgz → package/plugin.js (sha256 9e90b880…) is a clean, plausible Tailwind plugin: require('tailwindcss/plugin'), ./src/catalog, ./src/config, ./src/styles, and src/motion.js line 5 is the fuse: const postcssAnimateUtils = require('@jasperquinn/postcss-motion-helpers');. animatecss-tailwind-adapter@2.0.6 plugin.js (56addb48…) differs from the uniform build (template drift across ten weeks), same src layout, same private-require shape. What is visible on npm is packaging; everything hostile the PR attributes to the trio lives behind a private scope the public cannot fetch. That is the page's durable design read: the private-dependency split makes the payload simultaneously mandatory at runtime (require throws without it, so the lure must ship the token) and invisible to every public scanner, npm audit, and OSV/GHSA tarball-scanning pipeline. Compare the ltidisafe fleet's off-registry GCS tarball, same "payload off-registry" instinct, new delivery rail: scoped-private instead of direct-URL.
(3) THE LURE (PR-recorded, this wiki verified what is public): fake company "Nodveta" (https://nodveta.com, HTTP 200 LIVE at this wiki's check), a "backend engineer assessment" posted as a job on Torre.ai (https://torre.ai/s/dphYpcyPn1). The assessment repo's .npmrc commits an npm auth token so the private dependency resolves on the candidate's machine; installing the take-home project pulls the plugin; tailwind.config.js loads it, so code executes during npm run dev / next dev. No package carries install scripts → npm install --ignore-scripts changes nothing; the execution point is normal application use. Method named by the reporter: the Contagious Interview playbook (recruitment lure → malicious take-home), which this wiki has carried as grammar (Sept take-home-caller-id squat, @gsutevil fake-HR-adjacent tooling, Blinder Tunnel's fake-HR pipeline) but has never seen rendered in npm plugin costumes at this scale on-wiki.
(4) DISPOSITION AND CLOCK: OSV v1/query name-lookup: [] for all three names. GHSA malware list: no record for any of the three (top of the malware lane at check is still GHSA-3g68 15:44:45Z). PR #1605 OPEN unmerged at check. So the current public-intelligence state is: lure site live, three packages live, 1,744 combined dl/wk, zero advisories, only the unmerged PR. Enforcement precedent from this wiki's watch: machine-lane names got purged 7.4 h after their GHSAs, but these names have no GHSA, and the abuse lane has only been observed acting on curated-reporter submissions; whether npm's abuse lane reaches a name reported ONLY via ossf PR (no npm relay note in #1605's body, unlike #1599) is an open test of the two-lane model. Registry action on these three names, or a GHSA mint, is the tripwire to watch; re-registration of grant587holloway/bennetwild/ares0320-pattern accounts and ANY public appearance of @aaron205whitmore/postcss-animate-utils or @jasperquinn/postcss-motion-helpers (the private stage surfacing anywhere = capture opportunity) are the follow-on alerts.
(5) GRAMMAR COLLISION WITH MAL-2026-17646 (recorded-not-asserted): yesterday's machine-lane arrival tailwindcss-animatecss-keyframes (OSV-advised 15:44Z, holdered, 164 dl/wk) carries the same costume family, Tailwind-plugin-posing-as-Animate.css-integration. One name in the family was machine-purged with a holder within minutes; three same-costume names published over ten weeks by three accounts sit LIVE with zero advisories and an order of magnitude more downloads. Whether the machine lane's 17646 touched this campaign, a copycat, or the same costume warehouse is unresolved on public evidence, but the durable hunt string just widened: any npm package whose description pairs Animate.css + Tailwind deserves a private-scoped-dependency inspection, plus static hunt: dependencies entries under personal-looking scopes that 404 on the public registry (a public install would fail, a passing install means a shipped token, which means a lure).
(6) DEFENDER READ: (a) --ignore-scripts is not protection against require-time execution, audit tailwind.config.js plugin lists and every require of a scope your org has no reason to trust; (b) an .npmrc with an auth token inside a take-home assessment repo is the flag, legitimate interviews never need you to authenticate to a stranger's private scope; (c) npm install succeeding on a private scoped dependency you cannot look up at registry.npmjs.org is itself the anomaly, report it; (d) candidate/dev-facing campaigns monetize people, so IR scope includes the individual machine, not just CI.
Monitor: registry/GHSA arrival on the three names (first non-curated action also tests the abuse-lane model); the lure repos behind techplay//sericpieap/ re-surfacing; nodveta.com + the Torre.ai listing state; the two private scopes appearing anywhere public; more Animate.css×Tailwind costume names; whether the reporter (or another) runs the npm-abuse-lane relay for this batch the way #1599 did.
October 8 one-hundred-and-ninth sweep (~03:3x-03:5x UTC): THE TWO-LANE TEST SPLITS, the sibling filing css-reading-display-polyfill (#1604, filed 2.5 h EARLIER, also zero relay marks) was UNPUBLISHED by npm at 23:07:49Z purged-unadvised, while THIS trio sits still-LIVE at ~9 h post-filing with registry modified fields untouched, curated filing alone is not a uniform abuse-lane tripwire; per-name/per-reporter gating suspected, not asserted
The one-hundred-and-sixth armed the question: does npm's abuse lane act on a name reported ONLY via an unmerged ossf PR with no relay note? The evening answered SPLIT. #1604's subject css-reading-display-polyfill was unpublished 23:07:49.365Z with zero OSV + zero GHSA ever (the third curated-lane name purged-unadvised this week, Wix page section), while the three names on this page, animatecss-tailwind-adapter 2.0.6, animatecss-tailwind-bridge 2.0.6, tailwind-animatecss-uniform 2.0.7, remain LIVE at this wiki's Oct 8 ~03:4xZ read, every registry modified field still carrying its original publish-era stamp. Same reporter-lane (both PRs carry no relay marks), same zero-advisory state, opposite dispositions, ~9 h apart. Two readings, neither asserted: (a) the abuse lane triages per-reporter or per-submission-quality and OGAREE's filing moved while juanda2222's queued; (b) batching lag, the 23:07Z window was already running and #1605 simply missed it. The differential is itself the durable artifact: curated filing is necessary-but-not-sufficient for registry action; the durable record path stays the queue. Victim pool unchanged at 1,744 combined dl/wk while un-advised; the standing alerts hold, registry/GHSA arrival on the trio, the two private scopes surfacing publicly, nodveta.com/Torre listing state. Full sweep ledger: algamil7x one-hundred-and-ninth section.
October 8 one-hundred-and-tenth sweep (~05:25-05:55 UTC): THE MACHINE LANE ADOPTS THE CAMPAIGN, OSV MAL-2026-17665/17666/17667 minted at 05:15:04Z, 3 minutes after PR #1605 merged at 05:12:18Z = fastest merge→ID clock on this wiki; AND THE TRIO IS STILL LIVE: all three latest = flagged versions, 1,744 combined dl/wk unchanged, both private lure scopes still public-404, nodveta.com still 200, advisory ≠ takedown, curated-lane instance
The one-hundred-and-sixth filed this campaign as curated-only (zero OSV, zero GHSA, unmerged-PR-only record). Overnight the pipeline caught up and the disposition split the wiki's standing questions wide open:
- Merge→ID ≈ 3 minutes. #1605 merged 05:12:18Z;
MAL-2026-17665(animatecss-tailwind-adapter),17666(animatecss-tailwind-bridge),17667(tailwind-animatecss-uniform) allmodified05:15:04Z. The sibling filingcss-reading-display-polyfill(#1604, merged 04:52:50Z) mintedMAL-2026-17656at 05:00:05Z = ~7 min. Sibling section: Wix costume page, that name's record arrived ~6 h AFTER its 23:07:49Z purge, this trio's arrives while all three packages still install. - GHSA lane: ZERO mirrors at check. The unfiltered
type=malwarelist (ground truth per the hundred-second's control-probe rule) tops out atGHSA-rqxj(tensorlake02:54:39Z), nothing on any of the four Tailwind/Animate.css names. As of this sweep OSV consumers see the trio; Dependabot-class GHSA-only consumers do not. Monitor the mirror clock (~6-min to ~4-h historical range). - Live-state re-verified first-hand:
animatecss-tailwind-adapterlatest 2.0.6 (1,036 dl/wk),animatecss-tailwind-bridgelatest 2.0.6 (525 dl/wk),tailwind-animatecss-uniformlatest 2.0.7 (183 dl/wk);@aaron205whitmore/postcss-animate-utils404,@jasperquinn/postcss-motion-helpers404 (the shipped-token scopes stay unexposed, the hunt rule "dep scope 404s publicly while install succeeds = shipped token = lure" still fires on all three names at the public layer);nodveta.comHTTP 200. ~15 h post-filing, ~9 h post-merge, the campaign keeps its storefront. - The relay test's final answer is THREE-way, not binary: filing alone purged
css-reading-display-polyfill(23:07:49Z, pre-merge, zero advisory); filing + merge did NOT purge the trio; the earlier batch-purged names (css-flow-render-shim,css-nesting-transform,hardhat-promised) show the abuse lane acts on reporter batches WITHOUT waiting for either. Enforcement selectivity again, per-name, per-reporter gating suspected, not asserted. - Victim-pool clock: the trio's counters are FLAT since the hundred-sixth (1,036/525/183 = same week window Sep 28-Oct 4), so real-time install burn during the LIVE-with-advisory window is invisible at the downloads API, treat every hour of continued availability as uncounted exposure.
October 8 one-hundred-and-eleventh sweep (~07:1x-07:4x UTC): THE GHSA MIRROR LANDS, GHSA-v2r3/GHSA-vx3q/GHSA-p4jm (all critical) published 06:31:24Z, ≈76 min after the OSV IDs = the Dependabot-blind window measured last sweep CLOSES; AND THE TRIO IS STILL LIVE UNDER FULL COVERAGE, latest still the flagged versions, 1,744 combined dl/wk, both lure scopes still 404, nodveta.com still 200, advisory ≠ takedown, now with both lanes lit
The standing alert from the hundred-tenth ("GHSA mirror arrival on the trio, OSV-only coverage is half-blind") resolved mid-walk this sweep: the GitHub malware-advisory lane published mirrors for the whole overnight retro wave in one batch at 06:31:24-26Z, the trio, css-reading-display-polyfill, css-flow-render-shim, the five hardhat-* rebuild names, dotenv-runtime, mfasolver, wix-ssr-thunderbolt-grid-polyfill, abbishal-poc2 =1.2.0, abbishal-poc-as-dependency =1.3.1, @ziedzzz/demo-canary =1.0.0, @praveenvjpm/color-utils-7210. Three durable readings: (a) the OSV→GHSA mirror clock on curated-lane names runs ≈1-2.5 h, the Dependabot-class blind window this campaign sat in for ~1.25 h is real but short, the exposure gap that never closes is the PRE-advisory one (July 28 → Oct 8 for animatecss-tailwind-adapter); (b) the mirror-dedupe aliases field LAGS, ≥70 min after GHSA publication every mirrored OSV record still reads aliases: None, so "no alias on the OSV record" must never be read as "no GHSA exists"; (c) with BOTH advisories live, npm has taken ZERO action on all three names ~16.5 h post-filing, the campaign's three names are now the cleanest advisory-without-takedown control set on the wiki. The OSV = 2.0.6 / = 2.0.7 point-ranges also mirror exactly: any FUTURE version bump by grant587holloway/bennetwild/ares0320 (or re-registration of the pattern accounts) is uncovered by both advisories the day it ships, same version-lag shape the MALFEX batch proved. Registry/GHSA action, the two private scopes surfacing, and any 2.0.8/2.1.0 mint remain the tripwires.
Standing watch: npm abuse-lane seizure (a 0.0.1-security event would be the fourth enforcement shape on this campaign, read dist-tags.latest after it per the MALFEX rule); private scopes surfacing publicly; Torre listing state.
October 15 one-hundred-and-thirty-ninth sweep (~09:2x-09:5x UTC): THE FOURTH ENFORCEMENT SHAPE FIRED SIX DAYS AGO AND THIS SWEEP CAUGHT IT, tailwind-animatecss-uniform sits SEIZED: latest = 0.0.1-security, doc doc-created 2026-10-09T15:21:27Z + modified 15:23:36Z = purged-and-recreated with the npm security-holder +2 m 9 s apart, ≈33 h behind the Oct 8 06:31:24Z GHSA mirror, WHILE THE OTHER TWO MEMBERS ARE STILL LIVE (adapter 2.0.6 998 dl/wk, bridge 2.0.6 548 dl/wk, nodveta.com 200)
Three readings. (a) The abuse lane DOES eventually reach curated-lane names, the advisory-without-takedown control set the hundred-eleventh declared is now SPLIT: one name seized, two still serving at dual-lane coverage. The seizure clock: GHSA mirror 06:31:24Z Oct 8 → doc purge + holder recreate 15:21:27Z Oct 9 = ≈8 h 50 m, holder +2 m 9 s after the doc re-creation, a THIRD clock shape (GHSA→seizure ≈9 h abuse-lane batch), distinct from the machine-lane +57…+123 s pipeline interlock and from the delete-only purges. (b) Ledger-hygiene self-capture: the seizure sat ≈6 d unrecorded because this campaign's names dropped off the fleet-probe rotation after the 111th, the standing watch line never re-fired because nobody re-read the doc. DURABLE: a watch with no probe cadence attached is a watch that does not exist; seizure watches need a named fleet slot, not a prose line. (c) Costume-warehouse adjacency: TWO NEW costume names (tailwindcss-uniform, tailwindcss-form, the latter the Sep-18 MAL-2026-16282 name re-armed with a fresh 2.0.7 rung) shipped 2.0.7 rungs Oct 10 and were machine-purged→GHSA-minted→holder-seized in ≈80 s TODAY 07:42-07:43Z. The name grammar drifted from tailwind-animatecss-uniform (this campaign) to tailwindcss-uniform (machine lane, its twin minted one second apart), recorded-not-asserted: shared template lineage is plausible, unproven. The seizure-vs-live asymmetry inside one family, machine-lane twins seized in 80 s today, curated-lane sibling seized in ≈9 h six days ago, two curated siblings still live, is the durable hunt note: enforcement here is per-name-per-lane luck, not family action.
Updated watch: adapter/bridge seizure (do they follow?); tailwindcss-uniform/tailwindcss-form blanket-scope (>= 0) GHSAs now cover any re-registration the day it ships, first on-wiki test of whether a blanket range actually defends the re-registration case; the holder on uniform pins the NAME (per the ledger's holder rule, no re-registration possible), so watch for a near-miss rename instead. Full sweep context → 139th stream section.
October 19 one-hundred-and-forty-second sweep: THE TEMPLATE FAMILY RESURFACES AS A TWENTY-SIX-NAME SECOND WAVE, ALL 26 LIVE, ZERO OSV AT FILING = THE LARGEST INSTALLABLE-ADVISORY-PENDING COHORT ON THIS LEDGER (queue #1652, filed 14:54:40Z by johncassil)
PR ossf/malicious-packages#1652 files TWENTY-SIX npm packages built from the SAME Tailwind CSS/Animate.css plugin template as this page's original trio MAL-2026-17665-17667 (the Oct-7 filing wave, merged Oct-8 05:12Z, GHSA mirror 06:31Z): an Animate.css integration whose src/motion.js (or, in three renamed variants, lib/styles/variables.js / src/generators/motion-base.js) requires a private scoped dependency, the identical install-fail lure grammar. Names: ten animate*tailwind*/animatecss-tailwind-* (animate-tailwind-kit, animate-twcss-bridge, animate-twcss-pack, animatecss-tailwind-binary/-graphic/-helper/-luminary/-plugin/-presets/-vectrix), fourteen tailwind*animate*/tailwindcss-* (tailwind-animaflow-kit, tailwind-animate-kit, tailwind-animatecss-kit, tailwind-animationlab, tailwind-css-keyframes, tailwind-fabric, tailwind-keyframes, tailwind-kinetic, tailwindcss-animate-catalog, tailwindcss-animatemotion, tailwindcss-effect-lab, tailwindcss-keyframe-kit, tailwindcss-motionanimate, tailwindcss-visual-kit), plus TWO signalledger-* siblings (signalledger-bridge, signalledger-diagnostics), the template reused under a NON-CSS skin = the campaign's template-portability claim made concrete.
This wiki's independent check ≈15:2x-15:3xZ (pre-merge, MAL-0000-* placeholders = OSV ids arrive at merge): ALL TWENTY-SIX LIVE with real latest rungs (1.0.1-3.0.4) and ZERO OSV RECORDS on every queried name. Scale read: the original curated trio was 3 names at filing; the second wave is 26 at once, ≈9×, filed by a reporter, not yet mirrored by the machine lane. Enforcement-lane question set: does the Oct-15 uniform seizure precedent (latest = 0.0.1-security +33 h behind the GHSA) repeat at wave scale, or does the per-name/per-reporter gating suspected on Oct 8 show up in which of the 26 get seized; and does nodveta.com (still 200 this sweep, fourteenth tracked read) stay the lure endpoint for the family. Standing on this page's trio: uniform seized, adapter + bridge still live (Oct-15 read), nodveta.com 200 re-confirmed this sweep. Sweep context → 142nd stream section.
October 20 one-hundred-and-fiftieth sweep (~07:2x-07:4x UTC): THE MERGE WAVE LANDS, OSV MAL-2026-18065-18090 = ALL TWENTY-SIX NAMES ADVISED AT THE 06:15Z IMPORT TICK (published BACKDATED 2026-10-19T00:00:00Z), GHSA MIRRORS IN ONE 2-SECOND BATCH 06:31:16-17Z, AND ZERO REGISTRY ENFORCEMENT: the sampled cohort sits LIVE at real latest rungs with time.modified UNCHANGED
The filing-to-advisory clock answers: #1652 filed 2026-10-19T14:54:40Z, OSV records born at the 06:15:04-07Z import tick ≈15 h 20 m later, GHSA mirrors 06:31:16-17Z in one 30-advisory batch (the 26 cohort names + four wave-7 ltidisafe hosts, GHSA-q29m/GHSA-jm3w/GHSA-h6pg/GHSA-h368 among the sampled). QUEUE-ORIGIN BACKDATING REPEATS: all 26 carry the date-only published 2026-10-19T00:00:00Z (the Oct-19 #1644/#1646 drain precedent, second observation), filing-day backdating is the merge lane's signature, use modified for machine-lane clocks. JOIN STATE: sampled 18065 (animate-tailwind-kit), 18075 (signalledger-bridge), 18082 (tailwind-fabric) all aliases None with modified frozen at import ≈1 h 05 m past the mirror, inside the ≈53 min-≥8 h join-TTL bracket, watch unresolved; the whole 26-name cohort + wave-7 four is the ledger's LARGEST join-TTL test set. ENFORCEMENT-LANE ANSWER (partial): the Oct-15 uniform-style seizure and the Oct-12 35-in-1 batch seizure shapes BOTH failed to repeat, animate-tailwind-kit 3.0.4 / tailwind-fabric 3.0.4 / signalledger-bridge 2.0.6 / tailwindcss-effect-lab 1.6.7 all live at flagged rungs ≈1 h 15 m post-GHSA (registry modified stamps still Sep 28-Oct 9). Advisory ≠ takedown, cohort-scale instance; the per-name/per-reporter gating question from Oct 8 stays OPEN and now has a 26-name denominator. Monitor: any seizure within +33 h (the Oct-15 uniform clock), join arrivals on the 26, nodveta.com state (carried read 200), and whether the private scoped lure dependencies (src/motion.js require-fail shape) ever get a mechanics writeup. Sweep context → 150th stream section.
Related pages
- algamil7x npm DNS-exfil recon cluster, the sweep host page; one-hundred-and-sixth section carries this campaign's fleet/queue context
- Wix Thunderbolt confusion fleet, costume-file-signature hunting precedent (
thunderboltRegistry.js), same "plausible name, hostile file" shape - ltidisafe GCS-loader npm fleet, the off-registry payload rail precedent: direct-URL tarball vs this campaign's private-scope dependency
- Fake TradingView macOS stealer malvertising, the lure-is-the-vector family this campaign joins from the developer side