Overview / Settings

Settings

Manage your account, API keys, team, and integrations.

Profile Information
Your name and contact details shown in reports and invoices.
AR
Evaluation Defaults
Default settings applied to all new evaluation runs.
Evaluations above this delta will be flagged as BLOCK_DEPLOYMENT
Token window for temporal variance (Signal 5)
API Keys
Keys for authenticating the LLMGuard SDK and REST API.
🔐
API keys never expire unless you revoke them. Store them in environment variables — never hardcode in source code. Keys are displayed once at creation; after that, only the prefix is shown.
Production Key
lg_sk_prod_••••••••••••••••••••••••••••••XR7K
Active
CI/CD Pipeline Key
lg_sk_ci_••••••••••••••••••••••••••••••9M2P
Active
SDK Environment Setup
Copy this snippet to set your API key before running evaluations.
bashCopy
export LLMGUARD_API_KEY="lg_sk_prod_••••••••XR7K"

# Or in Python:
import llmguard
guard = llmguard.Client(api_key="lg_sk_prod_••••••••XR7K")
Team Members
4 of 10 seats used on Growth plan.
4 / 10 seats · Upgrade for unlimited
AR
— You
—
Admin
PM
Priya Mehta
priya@ailegallab.io
Editor
RK
Rahul Kumar
rahul@ailegallab.io
Viewer
SE
Sofia E.
sofia@ailegallab.io
Editor
Pending Invitations
Invitations expire after 7 days.
?
varun@ailegallab.io
Invited 2 days ago · expires in 5 days
Pending
Current Plan

Growth

$999/month · Billed monthly · Renews May 1, 2026

Active
Next invoice: May 1
Amount: $999.00
Usage This Month
Evaluation runs 64,230 / 100,000
Overage: $0.012 per run · Est. overage this month: $0
Batch jobs 23 / unlimited
Plan Features
Monthly eval runs100,000
Domain breakdown✓ Included
Custom prompt suites✓ Included
GitHub Actions plugin✓ Included
Analytics history12 months
Team seats10
SupportSlack, 4h SLA
On-premises TTMEnterprise only
SOC 2 reportEnterprise only
Model Integrations
Connect your model providers for native signal extraction.
OpenAI
GPT-4o, GPT-4o mini, GPT-3.5 — native logprobs
Connected
vLLM
Self-hosted models via vLLM server — native logprobs
Not connected
Anthropic / Claude
Via proxy model — evaluation_method: "proxy"
Proxy mode
Google Gemini
Via proxy model — evaluation_method: "proxy"
Not connected
CI/CD Integrations
Automate hallucination checks in your deployment pipeline.
GitHub Actions
Block PRs when hallucination delta exceeds threshold
- uses: llmguard/action@v1
  with:
    threshold: 0.05
    action: fail-pipeline
Active
HuggingFace Hub
Trigger evaluations when new model versions are pushed to Hub
Not connected
Webhook Endpoints
POST requests sent when evaluations complete.
Webhooks are POST requests with Content-Type: application/json. Body includes: batch_id, status, hallucination_delta, recommendation, and full report URL.
URLEventsStatus
https://ci.ailegallab.io/llmguard/callback batch.complete Active
https://slack-webhook.io/llmguard-alerts block.issued Active
Webhook Payload Example
json
{
  "event": "batch.complete",
  "batch_id": "batch_a7f21c",
  "status": "complete",
  "hallucination_delta": 0.127,
  "recommendation": "block_deployment",
  "confidence_interval_95": [0.089, 0.165],
  "report_url": "https://app.llmguard.io/reports/batch_a7f21c",
  "generated_at": "2025-07-15T14:32:11Z"
}
Email Notifications
Choose which events trigger email alerts to your registered email address
Evaluation complete
Sent when any batch evaluation finishes
BLOCK recommendation issued
Sent when hallucination delta exceeds threshold
Usage at 80% of monthly limit
Early warning before overage kicks in
Weekly summary report
Overview of all evaluations run in the past week
New team member joins
When an invitation is accepted
Slack Notifications
Post evaluation results to a Slack channel.
Connected to #llm-evals in AI Legal Lab
Connected
Password
Change your account password.
Two-Factor Authentication
Add a second layer of security to your account.
Not enabled

Without 2FA, anyone with your password can access your account and API keys. We strongly recommend enabling it.

Active Sessions
Chrome · macOS Current
Mumbai, IN · Last activity 2 min ago
VS Code Extension
API key auth · Used 3h ago
⚠ Danger Zone
These actions are irreversible. Please be certain before proceeding.
Delete all evaluation history
Permanently delete all runs, reports, and analytics data. Cannot be undone.
Revoke all API keys
Immediately invalidates all API keys. Your SDK integrations will stop working.
Delete Account
Permanently delete your account, all data, and cancel your subscription.