Skip to content

Reference — API & Configuration

Technical reference for FowyldAI APIs, CLI, and configuration.

Core API Endpoints

Health & Status

Endpoint Method Description
/ping GET Lightweight liveness check — returns {"status": "ok"}
/health GET Version + loaded model list
/health/ready GET Deep readiness probe (K8s) — returns 503 if degraded
/health/live GET Liveness probe (PID check)
/deployment/info GET Environment, tier, GPU info
/metrics GET Uptime, request count, latency, GPU stats
/metrics/prometheus GET Prometheus scrape format
/status/subsystems GET All subsystem health

Inference & Reasoning

Endpoint Method Description
/auto POST Auto-route to best model via Sovereign Brain
/classify POST Task type classification
/reason/light POST phi3-mini reasoning
/reason/deep POST mistral-7b reasoning
/reason/heavy POST openhermes-7b reasoning
/reason/light/stream POST SSE streaming (phi3)
/reason/deep/stream POST SSE streaming (mistral)
/reason/heavy/stream POST SSE streaming (openhermes)
/image POST Image generation (SD 1.5/SDXL)

OpenAI-Compatible (/v1)

Drop-in replacements for OpenAI API calls:

Endpoint Method Description
/v1/chat/completions POST Chat completions
/v1/embeddings POST Embeddings
/v1/models GET Model listing
/v1/images/generations POST Image generation
/v1/summarize POST Summarization
/v1/estimate POST Token estimation

Sovereign Brain

Endpoint Method Description
/brain/reason POST Primary intelligence endpoint
/brain/plan POST Goal decomposition
/brain/remember POST Store to institutional memory
/brain/recall POST Semantic memory search
/brain/compress POST Extractive summarization
/brain/cached_answer POST Semantic cache lookup

Agent System

Endpoint Method Description
/agent/run POST Agent execution
/agent/council POST Multi-model deliberation
/agent/tools GET Available tools list
/v1/agent/run POST Governed multi-step agent run

Enterprise

Prefix Description
/compliance/* GDPR, AI Act, SOC 2, FTC compliance checks
/billing/* Usage metering, tier status
/admin/* SOC2 readiness, incident runbook
/auth/sso/* SSO login, callback, userinfo
/knowledge/* Document scan, query, stats
/research/* Web search, deep research
/batch/* Batch job scheduler
/stream/* SSE event subscriptions
/interop/* MCP server/client, A2A protocol
/peer/* Cluster heartbeat, delegation, consensus

Interactive API docs

Swagger UI is available at http://127.0.0.1:8400/docs and ReDoc at http://127.0.0.1:8400/redoc when the Crown Engine is running in dev mode.


Configuration Files

All configs live in the config/ directory (YAML format):

File Purpose
config/dev.yaml App config: host, port, logging, inference, safety, brain
config/prod.yaml Production overrides
config/models.yaml Model registry: HF repos, VRAM requirements, quantization, node assignment
config/warm_pool.yaml Models preloaded in GPU VRAM at startup
config/security.yaml Rate limiting, CORS origins, alerting, encryption
config/billing.yaml Metering and tier definitions
config/gateway.yaml Gateway routing rules
config/clustering.yaml Multi-node cluster config
config/sovereign_brain.yaml Brain reasoning configuration
config/doctrine.yaml Governance doctrine rules
config/retention.yaml Data lifecycle and purge policies
config/storage.yaml Disk budget enforcement
config/trade.yaml Trading Post (cloud AI provider fallback)
config/embeddings.yaml Embedding model config

Environment variables are loaded from .env (see .env.example for all options).


CLI Commands

fowyldai

Starts the Crown Engine server. Equivalent to python -m uvicorn src.main:app.

fowyldctl — Customer Admin CLI

fowyldctl [--host http://127.0.0.1:8400] [--format text|json] [--quiet] <command>
Command Description
fowyldctl setup Guided setup wizard
fowyldctl models list List available models
fowyldctl models status Model load status and VRAM usage
fowyldctl models enable <name> Enable a model
fowyldctl models disable <name> Disable a model
fowyldctl models families Show model family intelligence brief
fowyldctl config show Display current configuration
fowyldctl config validate Validate config files
fowyldctl config reload Hot-reload configuration
fowyldctl config set <key> <value> Update a config value
fowyldctl security show Display security config
fowyldctl security rate-limit Configure rate limiting
fowyldctl security cors Configure CORS
fowyldctl security audit Run security audit
fowyldctl storage status Disk usage and budget
fowyldctl storage set-budget Set disk budget
fowyldctl safety show Safety policy status
fowyldctl safety set-policy Update safety policy
fowyldctl safety test Test safety filters
fowyldctl diag topology Show cluster topology
fowyldctl diag health Deep health check
fowyldctl diag delegation-log View delegation history
fowyldctl mesh status Cluster mesh status
fowyldctl mesh add-peer Add a peer node
fowyldctl mesh list-peers List peer nodes
fowyldctl compliance Compliance operations

fowyld-admin

Internal admin operations script (scripts/fowyld_admin.py).


Model Registry

Core models configured in config/models.yaml:

Model Provider Role VRAM Quantization
qwen25-1b Alibaba Classification (Scout) 3 GB
phi3-mini Microsoft Light reasoning 7.6 GB
mistral-7b Mistral AI Deep reasoning 4.5 GB GPTQ
openhermes-7b Nous Research Heavy reasoning 4.5 GB GPTQ
llama31-8b Meta Sovereign reasoning 16 GB
gemma-2b Google Fast reasoning
embedding sentence-transformers RAG embeddings
sd15 / sdxl Stability AI Image generation

Download models: python scripts/download_models.py [model_name]