Module daemon 

Smart Tree Daemon - System-wide AI Context Service

Runs smart-tree as a persistent background service that any AI can connect to. Provides:

• HTTP API for context queries
• WebSocket for real-time updates
• Foken GPU credit tracking
• HTTP MCP - Full MCP protocol over HTTP (not just stdio!)
• LLM Proxy - Unified interface to multiple AI providers with memory!
• Collaboration Station - Multi-AI real-time collaboration with Hot Tub mode! 🛁
• The Custodian - Watches all operations for suspicious patterns 🧹
• GitHub Auth - OAuth for i1.is/aye.is identity

“The always-on brain for your system!” - Cheet

Architecture

All AI features route through the daemon for persistent memory and unified state. The LLM proxy (OpenAI-compatible at /v1/chat/completions) is integrated directly. Collaboration hub enables humans and AIs to work together in real-time. The Custodian monitors all MCP operations for data exfiltration and supply chain attacks.

Structs

CreditTracker

Credit tracker for Foken earnings

DaemonConfig

Daemon configuration

DaemonState

Daemon state - The unified AI brain

DirectoryInfo

ProjectInfo

SystemContext

System-wide context

Transaction

Functions

load_all_tokens

Load all available valid tokens (for servers to accept any valid local token).

load_or_create_token

Load or generate the daemon auth token. Creates a new random token on first run and persists it.

load_token

Load existing token (for clients). Returns None if no token file exists. Prioritizes the system-level daemon token.

start_daemon

Start the daemon server

token_path

Get the path to the daemon auth token file. Respects ST_TOKEN_PATH env var (for systemd StateDirectory), falls back to ~/.st/daemon.token