Budgets & analytics
Keep AI review spend under control with fine-grained budgets, then use the analytics dashboard to see exactly where your budget goes — by team, repo, model, and more.
1. Fine-grained budgets
Every review agent gets fine-grained budgets in three windows: daily, weekly, and monthly — all denominated in USD. PURA tracks estimated spend per review and stops before a budget is crossed, so you never wake up to a surprise bill.
- Daily cap — resets every day at midnight UTC.
- Weekly cap — resets every Monday (UTC).
- Monthly cap — resets on the 1st of each month (UTC).
Caps are estimates, not guarantees.PURA tracks spend based on the provider's reported token usage. Set caps conservatively if you need a strict ceiling.
Separate from your plan's review allowance. Dollar caps are independent of the number of reviews your plan includes each month. If you use up that review allowance, PURA also pauses reviews until the 1st of the next month — see pricing.
2. Where caps apply
Caps can be set at three levels, from broad to specific:
- Agent level — a single cap covering all reviews that run through the agent.
- Scope level— a cap on a specific team, repo, author, or PR group you've defined. Use this to cap a noisy repo separately from the rest, or give one team a smaller monthly budget than the agent as a whole.
- Provider level — a cap on an individual AI provider within the agent, so you can limit how much goes to any one vendor.
More specific caps take effect independently of broader ones — all active caps are checked before a review runs.
Tip: scope-level caps are a great way to keep high-volume or experimental repos from consuming your whole agent budget.
3. What happens at the cap
When a budget window is exhausted, PURA doesn't silently cut corners — it follows a clear, predictable path:
- PURA checks the next provider in the agent's fallback chain to see if the review fits within its remaining budget.
- If a provider with headroom is found, the review runs there instead.
- If no provider fits within budget, the review is paused — skipped rather than overspent.
- Reviews resume automatically when the window resets (daily, Monday, or the 1st).
No silent overruns. PURA will never spend beyond your cap — it pauses instead. You can always trigger a new review once the window resets.
4. The analytics dashboard
The analytics dashboard gives you a clear picture of where your AI review spend goes. At a glance you'll see:
- Summary cards — total reviews, total spend, and total tokens consumed across your selected period.
- Budget cards — daily, weekly, and monthly spend compared against your caps, so you know exactly how much headroom you have left.
- Leaderboards — your top providers, models, repositories, users, and teams ranked by review volume.
Every view is filterable by date range, author, repo, model, provider, agent, or team — making it easy to drill into exactly which teams and repos are driving spend.
Sign in to see your data. The dashboard is available at app.pura.sh — log in with GitHub to access your organization's analytics.