Getting Started

Quick Start

Get InferexGuard — the enterprise AI gateway — running locally in under 10 minutes.

Prerequisites

Docker 24+ and Docker Compose v2
Git with submodule support
An OpenAI or Anthropic API key (at least one)

Installation

Clone the repo and initialise the Bifrost submodule:

bash

git clone https://github.com/covideoin/inferexguard.git
cd inferexguard
git submodule update --init --recursive

Copy and fill in your environment variables:

bash

cp .env.example .env
# Edit .env — add OPENAI_API_KEY and/or ANTHROPIC_API_KEY

Configuration

The .env file controls all enterprise plugin behaviour:

.env

# LLM providers
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...

# Postgres password (auto-generated in production)
POSTGRES_PASSWORD=changeme

# PII/DLP — detect | redact | block
PII_MODE=redact

# Data residency — hard-block non-compliant providers?
DATA_RESIDENCY_FAIL_CLOSED=false

# Compliance mode — soc2 | hipaa | gdpr
AUDIT_COMPLIANCE_MODE=soc2

Start the stack

bash

make dev
# or directly:
docker compose -f deploy/docker/docker-compose.dev.yml up --build

All six containers start: gateway, analytics, compliance, prompt-registry, postgres, redis.

bash

# Verify everything is healthy
curl http://localhost:8080/health        # gateway
curl http://localhost:8081/api/v1/health # analytics
curl http://localhost:8082/api/v1/health # compliance
curl http://localhost:8083/api/v1/health # prompt-registry

Enterprise Plugins

Three plugins run automatically in the gateway pipeline for every LLM request:

pii-dlp

Scans and redacts SSN, credit cards, emails, API keys before the request reaches the LLM. Set PII_MODE=block to reject requests outright.

bash

# Test PII redaction
curl http://localhost:8080/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o","messages":[
    {"role":"user","content":"My SSN is 123-45-6789, help me."}
  ]}'
# → prompt sent to OpenAI: "My SSN is [REDACTED:SSN], help me."

data-residency

Tag requests with X-Data-Residency: EU to restrict routing to EU-only providers.

bash

curl http://localhost:8080/v1/chat/completions \
  -H "X-Data-Residency: EU" \
  -d '...'

audit-log

Every request writes a chained audit record to PostgreSQL. SHA-256 prompt/response hashes, token counts, latency, and user identity.

bash

# View audit records directly
docker exec -it docker-postgres-1 psql -U inferlinks -c \
  "SELECT request_id, provider, model, prompt_hash, status FROM audit_events LIMIT 5;"

Analytics API

bash

# Cost summary — grouped by team
GET http://localhost:8081/api/v1/cost/summary?period=30d&group_by=team_id

# Daily time-series
GET http://localhost:8081/api/v1/cost/timeseries?period=30d

# 90-day forecast
GET http://localhost:8081/api/v1/cost/forecast?horizon=90d

# Model comparison
GET http://localhost:8081/api/v1/models/compare?period=30d

# Chargeback report (finance-ready)
GET http://localhost:8081/api/v1/reports/chargeback?period=2026-01

Compliance Reports

bash

# Start a SOC2 Q1 report (async)
curl -X POST http://localhost:8082/api/v1/reports/generate \
  -H "Content-Type: application/json" \
  -d '{"framework":"soc2","period":"2026-Q1"}'

# Poll status
curl http://localhost:8082/api/v1/reports/{id}/status

# Download when ready
curl http://localhost:8082/api/v1/reports/{id}/download

Supported: soc2, hipaa, gdpr. Period formats: YYYY, YYYY-MM, YYYY-Qq.

Prompt Registry

bash

# Create a prompt
curl -X POST http://localhost:8083/api/v1/prompts \
  -d '{"name":"support-agent","description":"Customer support"}'

# Add a version
curl -X POST http://localhost:8083/api/v1/prompts/support-agent/versions \
  -d '{
    "version": "1.0.0",
    "template": "Hello {{name}}, how can I help with {{issue}}?",
    "variables": {
      "name": {"type":"string","required":true},
      "issue": {"type":"string","required":false,"default":"your inquiry"}
    }
  }'

# Deploy to production
curl -X PUT http://localhost:8083/api/v1/prompts/support-agent/deploy \
  -d '{"version_id":"...","environment":"production"}'

# Render
curl -X POST http://localhost:8083/api/v1/prompts/support-agent/render \
  -d '{"environment":"production","variables":{"name":"Alice","issue":"billing"}}'

Production Deploy

Use your own nginx as a reverse proxy with Docker running the services:

bash

docker compose -f deploy/docker/docker-compose.server.yml build
docker compose -f deploy/docker/docker-compose.server.yml up -d

See the full deploy/ directory for Nginx config, certbot setup, and the deploy.sh operations script.

What's next?

Pricing

Self-hosted is free. See team/enterprise options.

GitHub

Source code, issues, and discussions.