How-To Guides — Administer & Configure FowyldAI¶

Task-based guides for administrators, operators, and integrators.

Set Up with the Guided Wizard¶

The fastest way to configure a new FowyldAI instance:

fowyldctl setup

This walks you through host, port, model selection, safety policies, and security settings.

Manage Models¶

List available models¶

fowyldctl models list

Check model load status and VRAM usage¶

fowyldctl models status

Enable or disable a model¶

fowyldctl models enable mistral-7b
fowyldctl models disable gemma-2b

Download a specific model¶

python scripts/download_models.py phi3-mini

View model families¶

fowyldctl models families

Model configuration lives in config/models.yaml. Each entry defines the HuggingFace repo, local directory, VRAM requirement, quantization, and cluster node assignment.

Configure Security¶

View current security settings¶

fowyldctl security show

Set rate limiting¶

fowyldctl security rate-limit --rpm 100 --burst 20

Configure CORS origins¶

fowyldctl security cors --origins "https://your-app.com,https://admin.your-app.com"

Run a security audit¶

fowyldctl security audit

Security configuration lives in config/security.yaml.

Configure Safety Policies¶

View current safety settings¶

fowyldctl safety show

Update safety policy¶

fowyldctl safety set-policy --policy strict

Test safety filters¶

fowyldctl safety test --input "Test content here"

Monitor Health¶

Quick liveness check¶

curl http://127.0.0.1:8400/ping

Full health status¶

curl http://127.0.0.1:8400/health

Deep readiness probe (for K8s)¶

curl http://127.0.0.1:8400/health/ready

Standalone health check script¶

python scripts/healthcheck.py --host http://127.0.0.1:8400 --verbose

View metrics¶

curl http://127.0.0.1:8400/metrics
# Or Prometheus format:
curl http://127.0.0.1:8400/metrics/prometheus

Check all subsystem status¶

curl http://127.0.0.1:8400/status/subsystems

Manage Configuration¶

View current config¶

fowyldctl config show

Validate config files¶

fowyldctl config validate

Hot-reload config (no restart required)¶

fowyldctl config reload

Update a specific setting¶

fowyldctl config set inference.max_tokens 4096

Manage Storage¶

Check disk usage¶

fowyldctl storage status

Set disk budget¶

fowyldctl storage set-budget --max-gb 100

Storage budgets are enforced via config/storage.yaml.

Run Diagnostics¶

View cluster topology¶

fowyldctl diag topology

Deep health check across all subsystems¶

fowyldctl diag health

View delegation history¶

fowyldctl diag delegation-log

Deploy with Docker (Production)¶

# Build the Crown Edition image
docker build -t fowyldai:crown --build-arg FOWYLD_EDITION=crown .

# Start the production stack
docker compose -f docker-compose.prod.yml up -d

The production Docker configuration runs with:

Read-only root filesystem
All Linux capabilities dropped
Non-root user
No-new-privileges flag
Port 8400 bound to 127.0.0.1 only (use the gateway for external access)

Set Up Peer Mesh (Multi-Node)¶

Check mesh status¶

fowyldctl mesh status

Add a peer node¶

fowyldctl mesh add-peer --host 192.168.1.50 --port 8400

List peers¶

fowyldctl mesh list-peers

Cluster configuration lives in config/clustering.yaml. Crown-to-Crown mTLS peering is configured in config/crown_tunnel.yaml.