Skip to main content
1

Create a virtual key

Sign in to app.wardin.ai → Console → Keys & Limits → Create key.Give it a name (e.g. dev-personal), set a monthly budget, and click Create. Copy the key — it starts with wardin_sk_.
2

Point your SDK at the gateway

Replace your provider base URL and auth token with the Wardin equivalents.
curl https://gw.wardin.ai/v1/messages \
  -H "Authorization: Bearer wardin_sk_YOUR_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "content-type: application/json" \
  -d '{
    "model": "claude-opus-4-8",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello, Wardin!"}]
  }'
3

Watch it in the dashboard

Open INGRESS in the Wardin dashboard. Your request appears in the live stream with token counts, cost, latency, and the model used — click it to open the signed receipt.If the request was blocked by a budget or policy, the POLICY feed shows the decision and the structured JSON error returned to your client. Your own usage, grouped into sessions, lives under MY LANE.

​
What just happened

Wardin processed your request through its enforcement pipeline:
  1. Auth — verified wardin_sk_ maps to a real key in your tenant
  2. Budget check — atomically checked Redis; would have returned a 429 if over limit
  3. Policy check — scanned the prompt against your active policies
  4. Cache lookup — checked for an exact-match response (free on a hit)
  5. Forward — sent the request to Anthropic with your real API key
  6. Cost commit — parsed token usage and committed spend back to Redis
  7. Async emit — sent the usage event to ClickHouse for dashboard analytics

​
Next steps

Connect Claude Code

Route all Claude Code traffic through Wardin for team cost tracking.

Set team budgets

Create per-team keys with monthly spending limits and alerts.

Configure policies

Add model allowlists, PII redaction, and injection detection.

API Reference

Manage everything programmatically via the REST API.