Token accounting for coding agents
See inside your software factory.
Maestro analyzes your org's coding agent sessions to show where tokens go, how your teams work with agents, and what to improve.
See Maestro in actionWorks with:Claude CodeCodexCursor
Your AI bill this month (and counting)
Maestro shows you where it goes.
Trusted by engineering leaders at:
Adoption
Understand adoption, team by team.
Follow what engineers delegate, which tools they use, and how their agent work reaches PRs, week by week. Spot where adoption is taking hold and where your rollout needs support.
| Personteam · primary tool | Est. spend8 weeks | Leading modelshare of spend | Top skilluses | PRsspend per PR |
|---|---|---|---|---|
 Mira OkaforGrowth · Claude Codefeatures, tests | $2,140Estimated spend twice a week over 8 weeks: $60, $75, $85, $100, $110, $130, $140, $30, $20, $150, $165, $180, $190, $205, $220, $280. | Claude Opus 5.564% of spend | verify-change22 uses | 34$63 / PR |
 Dev RamanGrowth · Codextests, refactors | $1,620Estimated spend twice a week over 8 weeks: $40, $25, $45, $30, $50, $35, $120, $90, $140, $105, $150, $115, $165, $125, $180, $205. | GPT-6 Astra71% of spend | write-tests18 uses | 36$45 / PR |
 Lena OrtizPayments · Cursorbug fixes, docs | $1,110Estimated spend twice a week over 8 weeks: $0, $0, $4, $0, $8, $30, $70, $120, $170, $140, $105, $95, $100, $90, $95, $83. | Claude Sonnet 5.558% of spend | triage-bug11 uses | 14$79 / PR |
 Sam WhitlockPlatform · Claude Coderefactors, chores | $950Estimated spend twice a week over 8 weeks: $70, $45, $80, $50, $35, $75, $60, $85, $40, $65, $55, $80, $45, $70, $50, $45. | Claude Opus 5.582% of spend | plan-feature9 uses | 9$106 / PR |
Spend and ROI
Know what your tokens buy.
Trace token usage and estimated cost across coding tools to initiatives, teams, and linked PRs. See how much of your spend turns into PRs, and how much never reaches one.
| Spend by repo | Relative spend | Est. spend | PRs | Cost / PR |
|---|---|---|---|---|
web-app | $6,420 | 88 | $73 | |
api | $4,980 | 61 | $82 | |
billing-service | $2,760 | 31 | $89 | |
mobile | $2,190 | 16 | $137 | |
infra | $2,046 | 10 | $205 | |
| Total | $18,396 | 206 | $89 |
Model choices
Match the model to the task.
Map model choices to real tasks: which models engineers reach for, where expensive ones dominate, and where a lighter model is worth testing.
Review
Claude Opus 5.578% of tokensClaude Sonnet 5.58% of tokensGPT-6 Astra5% of tokensGPT-6 Sol4% of tokensGPT-5.6 Sol3% of tokensClaude Haiku 4.52% of tokens
Skills and workflows
Find the friction in your workflows.
Track adoption of shared skills and see which instructions agents reference. Catch expected skills that go unused, see what each skill costs, and find where corrections and rework pile up.
- ImplementMade invitation links sign the user in and land on the workspace.
- TestfailedThe expired-link case still signed the user in.
- Skilltriage-bugTraced the failure to invitation tokens that never expire.
- CorrectionAdded an expiry check before an invitation is accepted.
- Reviewissues foundA reused link could still be accepted once after it expired.
- CorrectionRejected reused tokens at the same check as expired ones.
- DeliverOpened the pull request with the fix and three new tests.
For developers
Improve how you work with agents.
Weekly Rewind reviews your agent sessions, spots recurring patterns, and suggests specific changes to your prompts, instructions, and workflows. Every tip links back to examples from your own work.
Name the edge cases in your first prompt.
In 4 of your 9 feature sessions this week, the first test failed on a case your prompt never mentioned. Those sessions needed 2.5× as many corrections.
Make invitation links sign the user in and land on the workspace.
Make invitation links sign the user in and land on the workspace. Reject expired and reused links.
For platform and DevEx teams
Measure every change to your harness.
Your harness is the prompts, skills, and models your agents work with. Change one, compare rework and estimated cost across runs, then review linked PRs to see whether the change produced better code with less waste.
Security and compliance
Your data stays yours.
Every session, token, and insight Maestro handles is protected by enterprise-grade safeguards and transparent compliance practices.
- Zero AI training on your data, contractually guaranteed.
- SOC 2 Type II audited, with the report available in our Trust Center.
- Encrypted in transit and at rest.
- Single sign-on (SSO) and granular access controls.