Vergilant: monitoring for production LLM calls
Skip to contentVergilant<br>Sign inStart monitoring
14:03 · the setup<br>One URL change. That’s the integration.<br>Swap the base URL and add one header.<br>Anthropic and OpenAI, streaming included.<br>Responses come back untouched. Your API key passes through, never stored.
The whole integrationCopyfrom anthropic import Anthropic client = Anthropic( base_url="https://api.anthropic.com",+ base_url="https://proxy.vergilant.dev/anthropic",+ default_headers={"X-Monitor-Key": "lm_…"},) # Done. Every call is now watched.
14:05 · watching<br>Alerting isn’t a feature here. It’s the whole product.<br>No dashboards to build, no evals to configure. Two tripwires armed at signup, a third whenever you want it.<br>modellatencytokenscoststatusrequest body<br>Two armed at signup, one opt-in<br>error_ratearmedPast 10% failures in 15 minutes. The ping says why: “mostly 429 rate_limit.”<br>cost_spikearmed5× your 7-day baseline, learned automatically. No budgets to guess, no thresholds to tune.<br>silenceopt-inFor when traffic just stops. No error log will ever show it.
Meanwhile, in myapp-prod<br>14:04:12claude-sonnet-5 · 200 · 1.4s$0.01<br>14:04:51gpt-4o · 200 · 0.8s$0.02<br>14:05:22claude-sonnet-5 · 200 · 1.1s$0.09<br>14:05:58claude-sonnet-5 · 200 · 0.9s$0.31
nothing to report
Your agent is stuck in a loop.<br>Spend is 5.2× your baseline and climbing.<br>7-day baseline<br>last 15 min
VergilantAPPjust now<br>Cost spike: myapp-prod<br>Spend is 5.2× your 15-minute baseline. Looks like a loop.<br>$1.84 last 15 min · $0.35 typical
Caught four minutes in, with zero config. No budgets to guess: Vergilant learned this project’s baseline on its own.<br>Start monitoring<br>Sixty seconds to set up. Free tier included. No card required.
14:11:04 · resolved<br>Total damage: $1.84.<br>Sixty seconds of setup, paid for itself in four minutes. Back to work.
We store the envelope, never the letter.<br>There is no code path that writes a request or response body anywhere. The proxy is open source, so you don’t have to take our word for it.<br>What we see<br>Model name claude-sonnet-5<br>Latency total & time-to-first-token<br>Token counts input / output<br>Estimated cost<br>Status codes 200 · 429 · 500
What we never see<br>Your prompts<br>Model responses<br>Your users’ data<br>Never stored, never logged. Verifiable in the open-source proxy.
Free to start. A non-decision to keep.<br>Priced per project, not per seat. One caught loop pays for years.<br>Free<br>$0<br>The full experience, at hobby scale.<br>1 project<br>5,000 requests / month<br>All three alert types<br>Discord & email alerts<br>7-day metadata retention<br>Start free<br>Pro<br>MonthlyYearly 2 months free<br>$19.99 / month
Up to 10 projects<br>No request caps<br>90-day retention<br>Monthly cost reports<br>Multiple alert channels per project<br>Spend budgets<br>Start monitoring
Hit the free cap? That usually means real traffic. Upgrading is one click, and nothing about your setup changes.
Back to quiet. That’s the whole product.<br>Start monitoringJoin the Discord
© 2026 Vergilant