Overview
Real-time view of your engine, credits and last verdicts.
Recent activity last 8 calls · auto-refresh 30s
| Time | Verdict | Command | Cost |
|---|---|---|---|
|
⏳
Loading activity…Fetching the latest verdicts from the engine. | |||
Verdict legend
ALLOW safe to execute · fast path or sandbox-cleared.
WARN ambiguous outcome — recommend human review.
BLOCK simulation crossed the separatrix into Basin B.
Need more credits or a higher plan?
For now, top-ups and plan changes are handled by hand. Message Keter Labs on LinkedIn — we respond within 24 hours and can rotate, extend or upgrade your key in minutes.
Contact Keter Labs on LinkedIn →Usage
Credit consumption, top-up history and verdict distribution.
Top-up history
| When | Amount | Note |
|---|---|---|
📋No top-ups recorded | ||
Verdict Distribution last 50 calls
ALLOW WARN BLOCK
⚡ Fast Path vs Full Engine
Read-only commands (ls, cat, grep…) execute at 1 credit. Risky or ambiguous commands run the full engine at 5 credits.
🚨 Hard Cap Behavior
When credits hit zero, the API returns HTTP 402 until your account is topped up. Agents pause gracefully — no silent bypasses.
Audit log
Every /v1/check your key has issued, cryptographically sealed.
Decisions
| Time | Verdict | Conf. | Command | Intent | Cost | Proof |
|---|---|---|---|---|---|---|
|
⏳
Loading audit log… | ||||||
Playground
Live-fire a check against your key. Consumes real credits.
Compose a check
Raw Simulation Response
Tip · prime intent matters
The engine compares the command against the user's stated goal. A plausible intent that doesn't match the action raises divergence and trips Layer II selection.
What happens on Approve / Block?
These buttons let you override on WARN verdicts. The override is logged locally for this session — the original cryptographic verdict stays in your audit log untouched.
FAQ
Common questions about credits, verdicts, integration and security.
Trial keys are issued by Reality Kernel. Please reach out to our team or contact support through the official portal to request a pilot key.
The fast path (read-only commands like ls, cat, grep) costs 1 credit. Anything that hits the full engine — risky writes, ambiguous intent, pipelines — costs 5 credits. Every /v1/check response returns the exact cost in credits_consumed.
WARN means the engine detected a suspicious trajectory but is not certain — it recommends a human (or your agent's reviewer) makes the call. BLOCK means the action's simulation crossed the separatrix into Basin B (irreversible damage, exfiltration or compromise) and should not run.
Yes — the Playground exposes Approve anyway and Confirm block buttons after a WARN. Overrides are logged separately and never modify the original cryptographic verdict, so your audit trail stays clean for compliance.
Wrap any system or shell action in a call to POST /v1/check with the command and the user's prime intent. If the verdict is ALLOW, run it. If WARN, escalate to a human. If BLOCK, abort. See the Integration Guide for full implementation details.
The fast path returns in under 1 ms. The full engine runs in 50–500 ms — fast enough to wrap any write operation. Read-heavy agents typically see ~95% fast-path hit rate, so the average overhead is negligible.
Each verdict is hashed with SHA-256 over the decision payload (command, intent, verdict, timestamp, evidence). The hash is appended to a tamper-evident audit log — if any entry is altered later, the chain breaks. This is what makes the audit trail court-grade.
The API returns HTTP 402 on every subsequent /v1/check. Your agents won't silently bypass the engine — they pause, surface the error, and wait for a top-up. Contact your systems administrator to refill.
Commands and intents are processed in-memory by the engine and written to your tamper-evident audit log only. There is no third-party LLM in the decision path — the core is deterministic pattern matching plus filesystem diffing. Self-hosted deployments are available for regulated buyers.
Yes — contact Reality Kernel support to rotate your key. The old key is invalidated immediately, and the new one inherits your existing credit balance and history.
Profile
Account details, key metadata, and session controls.
Account
| Name | — |
| — | |
| Company | — |
| Plan | — |
| Status | — |
| Created | — |
| Expires | — |
| API key | — |
Profile fields are managed by Reality Kernel. To request modifications, please contact your systems administrator or reach out through our official support channel.
Sign out everywhere on this device
Wipes the cached key from both localStorage and
sessionStorage. You'll need your key to sign back in.