Welcome to the Glyphoxa documentation. These guides cover how to set up, configure, extend, and operate Glyphoxa — a real-time voice AI framework for TTRPG NPCs.
New to Glyphoxa? Follow this path:
Getting Started Architecture Configuration Testing Document Description Getting Started Prerequisites, build, first run, development workflow Architecture System layers, data flow, key packages, latency budget Configuration Complete config field reference, hot-reload, provider options
Document Description Providers Provider interfaces, supported providers, adding new providers, resilience NPC Agents NPC definition, entities, campaigns, VTT import, hot context assembly Memory 3-layer memory system, PostgreSQL setup, transcript correction, session lifecycle
Document Description MCP Tools Tool system, built-in tools, building custom tools, budget tiers Audio Pipeline Audio flow, transports, VAD, engine types, mixer, engine comparison
Document Description Commands Discord slash commands, voice commands, puppet mode, dashboard Deployment Docker Compose, Kubernetes / Helm, binary modes, production checklist Multi-Tenant Gateway, admin API, tenant model, session orchestration, usage tracking Observability Metrics, Prometheus, Grafana dashboards, per-tenant labels, health endpoints, alerting
Document Description Testing Running tests, conventions, mocks, provider testing, integration tests Troubleshooting Build issues, provider issues, runtime debugging, diagnostic steps
These documents explain the why behind each subsystem — design decisions, rationale, and specifications. The guides above explain how to work with each subsystem.