Pricing

Two products, each priced per engineer. Pick one to see its plans.

Coding Agent Visibility. See where your agent spend goes, and whether it’s getting healthier.See how VibeCheck works

Free

For teams of up to 5 engineers

$0

Upgrade when you pass 5 engineers

Start free
  • Up to 5 engineers
  • Claude Code, Codex and Cursor sessions, synced as each one ends
  • Estimated cost for every session, priced from its own tokens
  • How each session went: planning, testing, review, rework and corrections
  • Your org's spend by model
  • Install with one command, or roll out through your MDM

Enterprise

For large organizations with custom needs

Custom

Priced for your organization

Contact sales
  • Everything in Pro, plus:
  • Single sign-on (SSO)
  • Dedicated support with a named contact
  • Hands-on rollout help, including MDM
  • Security review and a data processing agreement
  • A contract and invoicing that fit your process

How it works

  1. Install once per laptop

    One command adds a stop hook to Claude Code, Codex or Cursor and starts a small sync service.

  2. Sessions sync when they end

    The hook only nudges the sync service, which uploads each finished session in the background.

  3. Each session gets priced and read

    VibeCheck prices it from its tokens and notes how it went: planning, testing, review, rework, corrections.

  4. See where spend goes

    Spend rolls up by model, then by person, team and skill. On Mondays, each engineer gets a Weekly Rewind.

Trusted by engineering leaders at:

Frequently asked questions

Why not just use the Anthropic or OpenAI console?

Your console shows what you spent and who spent it. VibeCheck shows what it bought: every Claude Code, Codex and Cursor session with its engineer, repo, kind of work and how it went, rework included. On Pro you also see cost per authored PR, and each Monday every engineer gets one change to try. VibeCheck vs. the provider consoles.

Can I show our CFO where the AI bill is going, and whether it's paying off?

Yes, on Pro. Usage breaks spend down by person, team, model, kind of work and skill over any range you pick, with cost per authored PR next to it. Sort sessions by cost and open the most expensive one. You'll see who ran it, what it was for, and whether it planned, tested or needed rework.

What leaves the laptop, and is our code safe with you?

Your data is encrypted in transit and at rest, Maestro AI is SOC 2 Type II audited, and your data never trains AI. We upload each session's prompts, replies, tool calls and edits, minus bulk nothing reads. The app shows summaries, not transcripts. Raw session data is deleted on a schedule after a session goes inactive.

Will it slow down our engineers' agents?

No. The hook in Claude Code, Codex or Cursor just nudges a small background service and returns. It never waits on the network. That service uploads each finished session and retries if the laptop is offline. Install is one command, with no Node, Python or Homebrew.

How hard is it to roll out VibeCheck to the whole team?

Not hard. Push it to every managed laptop through your MDM, or have each engineer run one install command, with no Node, Python or Homebrew needed. Sessions start syncing as they finish, and you can remove it the same way.

How accurate is the cost, and will it match our invoice?

It's precise for comparing people, models and kinds of work, because each session is priced from its own tokens, cache included, at current published reference prices. These are estimates, not invoice totals, so negotiated discounts and subscription plans aren't applied. Bedrock and Anthropic-direct sessions are both supported. Unpriced models are flagged, never guessed.

Will engineers feel watched?

They get something back. On Pro, every Monday each engineer gets a Weekly Rewind on their own sessions: one change to try, the sessions behind it, and a prompt to paste into their agent. Pro can also keep summaries private to each engineer. Admins then still see cost, tags and correction counts, but not the summaries.

What do we have after the first week?

You'll see where the spend went. Sessions appear as they finish, each with its cost and how it went, and spend by model fills in from day one. On Pro, Usage also breaks spend down by person, team, kind of work and skill, the optional GitHub App adds your merged pull requests for cost per PR, and each engineer gets a first Weekly Rewind the Monday after their first week with sessions.

Who counts as a paid engineer?

Anyone whose Claude Code, Codex or Cursor sessions VibeCheck tracks. Managers and others who only read the reports don't count. VibeCheck is free for up to 5 engineers, with every session's cost, how it went and spend by model. Team views, Weekly Rewind and cost per PR come with Pro, which is $30 per engineer per month billed annually, or $36 month to month.