We're live on Product Hunt →

CLI & GitHub App · npm, yarn, pnpm

Stop reading changelogs.
Let AI fix what breaks.

CLI & GitHub App run scans and upgrades in your repos. Radar is the fleet view of dependency health—paste JSON free, or saved reports on the dashboard with Pro.

Dependabot and Renovate open the PR; you still face breaking changes, failed tests, and manual rollbacks. Upshift explains what breaks, suggests code fixes, runs your tests, and rolls back automatically if anything fails.

They hand you chores. We help you finish them—with a receipt.

npm install -g upshift-cli
Add to GitHub Get Started Star on GitHub
Open source Source & issues Free tier, no card What AI guarantees

Built in the open. Used where it counts.

MIT
Open source license · full source on GitHub
6
Ecosystems scanned · npm, yarn, pnpm, pip, gem, go.mod
0
Silent rewrites · your tests stay the gate
Free
To start · no card, 10 AI credits included

Dependency upgrades are the worst kind of work

  • Breaking changes hidden in long changelogs
  • Migrations scattered across docs and blog posts
  • Tests fail, rollbacks are manual
  • Teams delay updates and accumulate risk

For teams: install once on GitHub

Team leads rarely want “install this CLI on ten laptops.” The Upshift GitHub App adds scan-on-PR and optional comments with one org-wide install; engineers keep their normal workflow. Power users still use the CLI for upgrade, fix, and migrations.

Add to GitHub App setup guide

Guardrails, not autopilot

Upshift accelerates upgrades with AI—it does not silently rewrite your codebase. Your tests stay the gate; you stay in the loop.

  • Test-gated upgrades — upshift upgrade runs your test script; failures trigger automatic rollback.
  • Review before apply — upshift fix --dry-run shows diffs; major bumps can require prompt or webhook approval.
  • Optional outcomes — UPSHIFT_RECORD_OUTCOMES=1 logs upgrade results locally for your own analysis.

When it breaks & guardrails · Human-in-the-loop (HITL)

  • ↩️ Upgrade runs your tests; auto-rollback on failure
  • 🛡️ fix --dry-run + approval hooks before apply
  • 📡 Radar: fleet dependency health
  • 🤖 AI explains + suggests fixes (you review)
  • 📋 What CI guarantees vs AI best-effort (honest limits)
  • 🔍 npm, yarn, pnpm (+ scan breadth for Python, Ruby, Go)

See it in action

Click any demo to see the full workflow

Demos show the CLI workflow. On the Free tier, explain --ai and fix use starter credits; Pro and Team include unlimited AI on the hosted dashboard with an API key. See pricing.

Not just another dependency bot

They hand you a list of chores. We did your chores for you—here's the receipt. Upshift explains breaking changes and suggests the code fix; you review and ship.

Dependabot / Renovate

  • Opens PRs with version bumps
  • You read changelogs and migration docs
  • You find and fix breaking changes yourself
  • You run tests and roll back manually if it breaks

Upshift

  • AI explains breaking changes in plain English
  • AI suggests code fixes — not just version numbers
  • One command: upgrade → run your tests → auto-rollback if anything fails
  • CLI-first: scan, explain, fix, upgrade from your terminal

Dependabot and Renovate are trademarks of their respective owners. We use these names for comparative, informational purposes only under fair use. No endorsement or affiliation implied.

Upshift turns upgrades into a repeatable workflow

Scan → AI Explain → Upgrade → Test → Rollback

  1. Scan for outdated or vulnerable dependencies
  2. AI Explain breaking changes, code patterns, and migration steps
  3. Upgrade the dependency
  4. Run tests and validate
  5. Rollback automatically if anything fails

What you get today

  • Dependency scan (npm, yarn, pnpm)
  • Risk assessment (low / medium / high)
  • AI explanations of breaking changes
  • AI-generated code fixes for migrations
  • One-command upgrades with tests
  • Automatic rollback on failure
  • VS Code extension
  • GitHub Action for CI scanning

Radar

Central view of dependency health across all your repos. Try Radar free (paste reports) or get Radar Pro with persisted dashboard, history, and alerts (Pro / Team).

Coming next

  • Scheduled upgrades + PR automation
  • Deeper AI assist for non-Node ecosystems (beyond scan)

Pricing

Start free: unlimited scans, upgrades, the GitHub App, and starter AI credits. Pro ($12/mo) adds unlimited AI and Radar Pro. Team ($39/mo) adds org-wide Radar Pro, audit-oriented defaults, and up to 10 seats.

Full pricing & feature matrix Sign in to upgrade

Canonical numbers: pricing.json.

FAQ

How is Upshift different from Dependabot?
Dependabot opens PRs with version bumps; you still read changelogs and fix breaking changes by hand. Upshift explains why things break in plain English, suggests code fixes (not just version bumps), and runs your tests with automatic rollback if anything fails. Think: Dependabot tells you what to upgrade; Upshift does the explaining and the fixing.
How can I approve risky upgrades (human-in-the-loop)?
By default, major upgrades prompt for approval in the CLI. Use .upshiftrc.json (via upshift init) to set approval.mode (prompt / none) and approval.requireFor (e.g. major). Use -y to skip approval for a run. For webhooks and full HITL, see How we do human-in-the-loop (HITL).
Do credits expire?
No. Credits roll over forever.
What happens if tests fail?
Upshift restores your last working dependency state.
Can I use my own test command?
Yes, Upshift runs your existing npm test script.
Where are Privacy and Terms?
Privacy Policy and Terms of Service for upshiftai.dev and related services. Paid checkout may show additional Stripe or dashboard terms.
Refunds and billing help?
Contact support through your dashboard after sign-in; refunds follow the policy shown at purchase.

Upgrade dependencies without the stress

You can try Upshift free right now — no credit card required.

Add to GitHub Get started free Star on GitHub

Want product updates?