Skip to content
Vinkius

See everything your connections are doing. In real time.

Usage, costs, and performance for all connections your AI makes. All in one dashboard.

Try for Free · No credit card
Pillar 01

Telemetry Primitives

Eight immutable metrics per V8 endpoint. Request volume, latency, error thresholds, and active state transitions. Updated synchronously.

Pillar 02

Execution Cost Attribution

Deterministic token expenditure per V8 endpoint execution. Define inference baselines and track cryptographic governance savings per payload.

Pillar 03

Endpoint Telemetry

Per-endpoint latency and failure rates across the MCP infrastructure. Identify AST bottlenecks and state transition failures before they impact egress.

Metrics

Deterministic Observability.

Eight immutable metrics tracked per V8 endpoint. We expose raw latency, error boundaries, and connection fidelity directly from the routing engine. Absolute determinism.

Total Requests 12,847

Cumulative payload evaluations since endpoint initialization.

P95 Latency 142ms

95th percentile response latency. Evaluated on each synchronous transition.

Error Rate 0.3%

Percentage of aborted evaluations in the last 24 hours.

Avg Response 89ms

Mean response latency across all payload evaluations.

Highest Latency db_query

The V8 endpoint with the highest average evaluation latency.

Highest Volume search

Most frequently evaluated endpoint by authenticated clients.

Active Clients 7

Authenticated clients actively connected to this V8 instance.

Last Activity 2s ago

Time since the last authenticated payload evaluation.

Live Feed

Live Payload Feed

Real-time AST evaluation feed. Endpoint, execution latency, and deterministic status codes.

Live Stream

Payload evaluations are logged synchronously. Timestamp, endpoint, transition time, and HTTP status.

Mutation Detection

Each payload is deterministically tagged as QUERY, MUTATION, or DESTRUCTIVE. Strict visibility into AST write access.

Data Redaction Counter

How many cryptographic keys, PII, and restricted identifiers were intercepted per V8 egress.

Audit Trail

Immutable Execution Ledgers

Searchable ledger of each payload evaluation. Endpoint, transition status, latency, and interception counts.

All Calls Logged

Endpoint, method, state transition status, latency, interception counts, and client identity.

Retention by Plan

Free: real-time only. Team: 7 days. Business: 30 days of searchable history.

Status Codes

Green for deterministic success, amber for malformed client payloads, red for infrastructure failures. Spot execution bottlenecks synchronously.

Redaction Tracking

How many sensitive patterns were intercepted and redacted per execution egress.

Connections

Identity Access

Authenticated client connections visible in real time. Client signature, IP, and cryptographic session duration.

Connect / Disconnect

Which clients authenticated, when sessions terminated, and exact duration constraints.

Transport Type

Streamable HTTP or SSE. Deterministic visibility into the transport layer of each connected client.

Client Identity

Client signature, protocol version, and IP. Absolute visibility into who accesses your governed endpoints.

SIEM Streaming

SIEM Integration

Stream all execution events to Splunk, Datadog, or custom webhooks. Synchronous delivery.

Splunk

Provide your endpoint and token. Execution events stream to your SIEM via HEC in real-time.

Datadog

All Datadog regions supported, including GovCloud. Execution telemetry appears synchronously.

Custom Webhook

Stream execution events to any HTTPS endpoint. Cryptographically signed payloads with deterministic retries.

FinOps

FinOps and Attribution. No hidden LLM overhead.

Hardcap token costs at the server level. We attribute exact expenditure per AI agent tool call against your defined LLM rate.

$8.52 Estimated Cost

Total estimated expenditure based on execution token volume and your inference baseline.

$1.86 Savings

Tokens saved by cryptographic constraints. Inference budget your clients did not burn.

$0.002 Cost / Payload

Average deterministic cost per payload evaluation.

21.8% Savings Rate

Percentage of total inference budget preserved by Vinkius governance.

Daily Spend Chart

Daily expenditure vs savings. Visualize exactly where cryptographic governance reduces token bloat.

Cost by Server

Which V8 endpoints consume the most tokens. Exact cost vs savings per execution boundary.

Your LLM Rate

Configure your inference baseline per million tokens. All dashboards recalculate synchronously.

Tool Intelligence

Identify latency bottlenecks. Isolate transition failures.

Per-endpoint latency and failure rates across the entire MCP infrastructure. Isolate bottlenecks before they impact egress.

8,420 Evaluations

Total payload evaluations across all V8 instances.

24 Unique Endpoints

Distinct endpoints actively evaluated by connected clients.

187ms Avg Latency

Mean response latency across all V8 endpoints.

142 Total Errors

Aborted evaluations. Isolate which endpoints experience transition failures.

Slowest Tools Ranking

Endpoints ranked by evaluation latency. Identify which processes bottleneck the overall AST execution.

Most Failing Tools

Which V8 endpoints experience transition failures. Error rate ranking across all executions.

AST Health Matrix

Latency vs evaluation failure rate per endpoint. Green = deterministic, amber = degraded, red = critical fault.

Controls

Execution Overrides

Synchronous kill switches, forced connection termination, cryptographic privacy locks.

Synchronous Kill Switch

The V8 instance halts deterministically. Authenticated connections and AST evaluations are revoked synchronously.

Authenticated Clients

Real-time cryptographic distribution of connected AI agents. Cursor, Claude, Windsurf, or custom integrations.

Log Cryptography

Endpoints published on the Integration Registry strictly enforce privacy. Individual execution logs are blocked; aggregate telemetry only.

Terminate Sessions

Force-terminate all active authenticated sessions. Synchronous execution. Zero grace period.

FAQ

Analytics
on Vinkius.

Vinkius Analytics provides eight immutable metrics per V8 endpoint: request volume, P95 latency, error rate, average response time, highest-latency endpoint, highest-volume endpoint, active client count, and last activity timestamp. All metrics update synchronously on the Vinkius dashboard.
Vinkius attributes exact token expenditure per AI agent tool call against your defined LLM inference baseline. You configure your rate per million tokens, and Vinkius calculates cost, savings, and savings rate deterministically for every execution.
Vinkius streams execution events synchronously to Splunk via HEC, Datadog across all regions including GovCloud, and any custom HTTPS webhook endpoint. Payloads are cryptographically signed with deterministic retry delivery.
The Vinkius audit trail is an immutable execution ledger that records every payload evaluation: endpoint, method, state transition status, latency, interception counts, and client identity. Retention varies by plan — 3 days on Lite and Starter, 7 days on Pro, and 30 days on Business.
Yes. Vinkius provides a synchronous kill switch that halts the V8 instance deterministically. All authenticated connections and AST evaluations are revoked instantly. You can also force-terminate individual sessions with zero grace period from the Vinkius dashboard.
Vinkius Tool Intelligence ranks every MCP endpoint by latency and failure rate. It identifies the slowest and most failing endpoints across your infrastructure, and visualizes latency vs error rate in an AST Health Matrix so you can isolate bottlenecks before they impact egress.