Antworten rund um CostNerve

CostNerve brings Vercel, GitHub, OpenAI, Neon and Supabase costs into one project-level view. It helps you see what is spending, why cost changed, what is still exposed and where action is needed before a surprise bill grows.

The onboarding flow is designed to reach a first useful cost view in under five minutes when provider permissions and APIs respond normally. Large accounts or provider rate limits can take longer, and CostNerve shows when discovery is still partial.

The core product focuses on Vercel, GitHub, OpenAI, Neon and Supabase. Keeping the connector set focused makes cross-provider project mapping and cost explanations more useful than a shallow list of integrations.

Yes. Connections are read-only by default. Cost-control permissions are separate and are requested only if you explicitly choose automatic protection.

GitHub activity can help connect repositories, commits, workflows and deployments with projects and cost changes. It provides technical context; CostNerve does not treat correlation alone as proof that a commit caused a bill increase.

CostNerve marks discovery as partial instead of pretending the account was fully scanned. You can resolve the permission or rate-limit issue and continue discovery without losing your progress.

Automatic Project Map groups resources and usage from different providers around the product or project they belong to. You can confirm or correct mappings, and ambiguous spend remains Unallocated rather than being silently guessed.

Unallocated means there is not enough evidence to assign a cost confidently to a project. Keeping it visible prevents a neat-looking dashboard from hiding real spend.

Exact cost comes from provider billing evidence. Estimated cost is reconstructed from usage and pricing when exact billing data is not available. CostNerve labels the difference so you know how much confidence to place in each number.

Differences can come from billing delays, credits, taxes, pricing periods, currency conversion, partial discovery or estimated usage. Check the confidence label, sync status and Unallocated spend before comparing totals.

Explain My Bill connects a cost change with available technical evidence such as deployments, commits, model usage, database activity or resource changes. It shows the evidence and confidence rather than inventing a cause.

It can correlate Vercel cost changes with deployments and affected resources, then show the evidence that makes the deployment relevant. This helps you investigate quickly without claiming causation when the evidence is incomplete.

Yes. CostNerve is designed to surface sudden changes in AI spend and put model or usage signals beside the project that generated them, alongside the rest of the stack cost.

Yes. A traffic surge, release or runaway process can increase hosting, AI and database spend together. CostNerve groups those signals by project so one incident does not look like several unrelated bills.

They are progressive budget thresholds. They give you earlier warning as spend approaches the limit, while sudden-spend detection covers fast changes that a monthly threshold could catch too late.

A CRITICAL alert is reserved for conditions that need immediate attention, such as dangerous spend velocity or a protection event. The product is designed to make these alerts actionable rather than sending routine noise at the highest severity.

Runaway Spend Protection combines spend monitoring, verified critical notifications, controllability and optional cost controls. Its purpose is to reduce the amount of spend that can continue growing unnoticed.

Protection Score reflects economic coverage: how much spend can be monitored or controlled and how much remains exposed. A high-looking score should not hide meaningful uncontrolled spend.

Emergency Brake is an optional last-resort control for runaway cost. It is off by default and must be explicitly enabled. When available, actions are designed to be resource-specific, reversible and audited rather than a blanket shutdown of your stack.

Not by default. Connecting a provider does not enable automatic changes. You choose whether to enable cost controls and define the protection you want before any automated action is possible.

That is the intended safety model. Protection should target the resource or workload creating the exposure instead of disabling unrelated infrastructure. Available actions depend on provider capabilities.

Protection actions are designed to be reversible where the provider supports reversal. CostNerve also keeps an audit trail so you can see what happened and why.

It should alert and show the uncertainty rather than taking a broad destructive action. Ambiguous attribution remains visible until you confirm the mapping or stronger evidence arrives.

CostNerve could not exhaustively inspect everything required for a complete map. Common causes are missing permissions, provider rate limits, temporary API failures or timeouts. Follow the displayed blocker and retry discovery.

Check the connector status for revoked authorization, expired credentials, permission changes, provider outages or rate limits. Reconnect only the affected provider; your existing project mapping and history should remain visible.

Provider billing data is not always real time. CostNerve can detect some usage changes earlier than finalized billing, but the interface distinguishes current evidence from delayed or estimated cost.

Verify that notifications are enabled for the installed PWA, that the browser or operating system allows notifications, and that the test push succeeds. CostNerve should not describe protection as complete until the critical channel has been verified.

No for basic cost visibility. Installing the PWA is useful when you want a direct critical push channel. You can continue setup later without losing onboarding progress.

Yes where the skipped step does not create a safety problem. Guided setup preserves progress so you can connect another provider, resolve partial discovery or configure protection later.

Provider credentials are intended to remain server-side and should not be exposed in browser-visible data or logs. The product follows least-privilege access and separates read-only access from cost-control permissions.

No. Tenant isolation is a core security boundary. Administrative access is separately authorized and customer data should never be exposed across tenants.

Use the connection controls to revoke the provider integration. Revocation should stop future access while retaining the minimum historical information needed for your account, subject to the product retention policy.

It should not. Secrets are treated as sensitive credentials and must not be written into application logs or exposed in support diagnostics.

Project Economics compares attributable technical cost with project revenue so you can reason about technical margin, not just infrastructure spend. Revenue connection is optional and comes after core cost visibility.

No. Cost monitoring, attribution and protection do not depend on Stripe. Revenue data is an optional later step for Project Economics.

Start with the project showing the largest change, check sudden-spend signals, review Explain My Bill, then inspect provider and resource evidence. If the cost is still accelerating, use verified alerts and any explicitly enabled protection controls rather than making a blind account-wide change.

Check spend velocity, the project with the largest delta, recent deployments, API usage and database activity. Then separate exact from estimated cost and inspect Unallocated spend so a missing mapping does not hide the real driver.

Identify the affected project and API workload, verify whether retries, loops, traffic or model changes explain the increase, then apply the narrowest safe control available. Emergency controls should only be used after they have been explicitly enabled.

Use project-level budgets, 70/85/100 thresholds, sudden-spend detection, a verified critical notification channel and a reviewed protection plan. The goal is to shorten the time between abnormal spend starting and someone being able to act.

Yes. Cross-provider visibility is a core use case. A single product event can change hosting, AI and database costs, so CostNerve is designed to show them in one project context.