A service is one running application with the SDK installed, however many instances it runs on. Probes and captures are unlimited on every plan, because nothing about debugging should be metered.
First incident we work with you is free · Cancel any month · No seat counts · No host counts · No capture limits
Install the SDK and see a real capture the same afternoon.
For teams running real production traffic who want the whole stack instrumented, not one service.
For teams whose security review has to sign off before anything touches production.
Nothing about debugging is metered. Every probe is bounded by controls you set instead.
Every probe carries an expiry. Set it in seconds or minutes. When the clock runs out the probe clears itself, whether or not it ever fired.
Cap how often a probe is allowed to capture. High-traffic lines stay bounded no matter how much traffic they see.
Set the number of captures a probe is allowed. Once it has collected what you need, it clears itself.
On Enterprise your security team sets the ceilings on all three at the organisation level, so nobody can place a probe outside policy.
| Free | Professional | Enterprise | |
|---|---|---|---|
| Probes and captures | |||
| Probes | Unlimited | Unlimited | Unlimited |
| Captures | Unlimited | Unlimited | Unlimited |
| Probe TTL, rate limit, hit expiry | Included | Included | Included |
| Team-level probe policy | — | Included | Included |
| Org-level probe policy and ceilings | — | — | Included |
| Scope | |||
| Services | 1 | Unlimited | Unlimited |
| Team members | Unlimited | Unlimited | Unlimited |
| Capture history | 7 days | 30 days | Custom |
| Agent | |||
| MCP server for every coding agent | Included | Included | Included |
| All supported runtimes | Included | Included | Included |
| Deployment | |||
| Managed cloud | Included | Included | Included |
| Self-hosted or private VPC | — | On request | Included |
| Data residency | — | — | Included |
| Security and governance | |||
| Read-only by architecture | Included | Included | Included |
| Default PII redaction | Included | Included | Included |
| Immutable audit log | Included | Included | Included |
| Custom redaction rules per field and file | — | — | Included |
| Approval gates before a probe arms | — | — | Included |
| RBAC | — | Included | Included |
| Per-namespace probe policy | — | — | Included |
| SSO | — | Google, GitHub | SAML, SCIM |
| Support | |||
| Channel | Community | Shared Slack | Named engineer |
| Response | — | 1 business day | Contracted SLA |
| Security review support and DPA | — | — | Included |
One application with the SDK installed. Run it on forty pods and it is still one service. We do not bill on hosts, containers, traffic or seats.
No limit on any plan. You should never meet a paywall in the middle of an incident. We bill for how much of your stack is instrumented, not how hard you debug.
Every probe carries a TTL, a rate limit and a hit expiry that you set when you place it. It clears itself when any of the three is reached. On Enterprise your security team sets the ceilings on those values at the organisation level.
Because it is how you use the product. Your coding agent drives the probes through our MCP server, so gating it would leave the free plan with nothing to do.
Yes. Default redaction runs at the agent on every plan, before anything is captured. Enterprise adds custom rules per field and per file for your own sensitive data.
Probes are non-blocking and read-only. No thread pauses and no request waits on a capture. We measured under 1% overhead at 3,000 RPS on our reference service, and we will run the same benchmark against yours during the POC.
Yes, on Enterprise. Control plane and capture store both run in your VPC and no captured value leaves your network.
Node.js, TypeScript, Java, Kotlin, Scala and Python today. Go and Ruby are on the roadmap. Tell us what you run and we will tell you where you sit in the queue.
Paste one prompt into your coding agent, pointing it at our installation guide. The agent installs the SDK, wires your entrypoint, turns on source maps and registers the MCP server. The same agent that installs HyperProbe is the one that will use it. Most teams are capturing inside fifteen minutes.
Yes. The SDK is identical. Moving to self-hosted changes where the control plane runs, not your instrumentation.
30 minutes. Your service. Your incident. You will see a confirmed root cause before the call ends — or there is nothing more to discuss.
Node.js · TypeScript · Java · Python · Works with Cursor, Claude Code, Codex, Opencode