Pinned Tweet
One paste. Four agents. A lesson your team can take.
Start free → enablementstudio.ai
RyanTHawks retweeted
Dink Bot 🏓 a free Grok Bot for pickleball.
snap your club's score sheet, it reads every game, asks about the smudges, and spits out a DUPR-ready CSV. never judges your third shot drop.
@PPAtour @MajorLeaguePB @TheKitchenPB @Pickleball @USAPickleball
x.ai/bot/aO8T9h4gksIpO1Xb6zR…
This month alone: [email protected] shipped a credential worm on install (Oct 8). @subql/[email protected] got a swapped tarball through trusted publishing (Oct 5). Then Node's Brotli decoder got an RCE patch (CVE-2026-111497).
I got tired of hoping my agents wouldn't npm install into that, so they don't anymore.
Supply-Chain Guard freezes installs to lockfile versions, pins checksums, waits a week before trusting a brand-new release, and checks machines for stealer tricks (.pth hijacks, weird LaunchAgents, plaintext key paths). Read-only by default. Asks before it changes anything.
x.ai/bot/I8xJf7ubpVXxfDvf6BT…
the PR is green. cool.
did the agent actually keep the tests honest, or did it delete the asserts that were failing?
Test Cop reviews the diff and returns CLEAN, SUSPECT, or WEAKENED with file:line proof. it never merges. it never rewrites your suite.
x.ai/bot/-IP_HE8l0UOByWip4DB…
Ship Evidence Auditor
coding agents say the ship passed. this one checks the receipts (CI, verify, merge method, ship log) and pauses the lane when they don't.
plus a report-only Cloudflare security-audit for Workers/Pages, separate from ship audits.
x.ai/bot/5l-uALmS59Wb9hZAMvG…
RyanTHawks retweeted
Every week there's a new poisoned npm or PyPI package. I got tired of hoping my agents wouldn't install one, so they don't anymore.
Supply-Chain Guard is a free Grok Bot template. It freezes installs to the versions already in your lockfiles and pins checksums. It waits a week before trusting a brand-new release, and it checks your machines for known stealer tricks. It's read-only by default and asks before it changes anything.
Grab it here: x.ai/bot/I8xJf7ubpVXxfDvf6BT…
RyanTHawks retweeted
Your AI agent says "done."
Prove it.
Software Factory is one Grok Bot that takes a code change from brief to shipped, and it shows receipts at every step:
🔒 A state machine with a hash-chained ledger. No skipped steps, and every new commit gets re-verified.
🔍 An audit against your brief with file:line evidence, not vibes.
🧪 It catches quietly weakened tests and supply-chain drift.
✅ Live QA after deploy, plus a regression check when you swap models.
💸 Cheap models do the heavy lifting. If a router breaks, it fails loud instead of quietly running up your bill.
Nothing merges, deploys or posts without your yes.
Built on ideas from @poteto @mattpocockuk @kieranklaassen @obra @ericzakariasson 🙏
x.ai/bot/pHt42M6Br4qDLEFQ8bl…
Dink Bot 🏓 a free Grok Bot for pickleball.
snap your club's score sheet, it reads every game, asks about the smudges, and spits out a DUPR-ready CSV. never judges your third shot drop.
@PPAtour @MajorLeaguePB @TheKitchenPB @Pickleball @USAPickleball
x.ai/bot/aO8T9h4gksIpO1Xb6zR…
Your AI agent says "done."
Prove it.
Software Factory is one Grok Bot that takes a code change from brief to shipped, and it shows receipts at every step:
🔒 A state machine with a hash-chained ledger. No skipped steps, and every new commit gets re-verified.
🔍 An audit against your brief with file:line evidence, not vibes.
🧪 It catches quietly weakened tests and supply-chain drift.
✅ Live QA after deploy, plus a regression check when you swap models.
💸 Cheap models do the heavy lifting. If a router breaks, it fails loud instead of quietly running up your bill.
Nothing merges, deploys or posts without your yes.
Built on ideas from @poteto @mattpocockuk @kieranklaassen @obra @ericzakariasson 🙏
x.ai/bot/pHt42M6Br4qDLEFQ8bl…
Every week there's a new poisoned npm or PyPI package. I got tired of hoping my agents wouldn't install one, so they don't anymore.
Supply-Chain Guard is a free Grok Bot template. It freezes installs to the versions already in your lockfiles and pins checksums. It waits a week before trusting a brand-new release, and it checks your machines for known stealer tricks. It's read-only by default and asks before it changes anything.
Grab it here: x.ai/bot/I8xJf7ubpVXxfDvf6BT…
Most enablement content gets built, shipped, and forgotten. Enablement Studio is an AI suite for instructional designers: build a Role, coach a live call, get a Critic's review of your draft, or have the Creator write the lesson. Free to start. enablementstudio.ai
Most training requests aren't training problems.
"Our new reps take forever on billing tickets" usually means a missing job aid, a skipped step, or no access to the billing system. A course fixes none of that.
Enablement Studio starts there. Paste the gap, a job description or an SOP. Specialist agents sort out what training can actually fix, then draft the objectives, practice, a quiz, a 30-minute session and spaced follow-ups. Creator turns it into an online lesson your team takes on their own. You decide what to keep.
There's a free plan with no card, and you can try the example without an account.
enablementstudio.ai
My coding agents don't trust each other. That's the whole design.
Here's one real change, PR #224, traced from my ask to merged, with the receipts.
🤖 Made with AI
ALT Isometric pipeline of locked robot stations: Ask, Brief, Build, Review, CI 4,549 passed, Ship-log, Audit, Merged.
The rules, short version:
No agent grades its own work.
A different model family reviews every change.
Scripts refuse instead of trusting summaries.
Only I approve spend, sends, and deletes. Premium models go through my chief of staff.