Skip to content

configuration.md

Latest commit

 

History

History
100 lines (82 loc) · 3.52 KB

configuration.md

File metadata and controls

100 lines (82 loc) · 3.52 KB

Configuration

Environment Variables (.env)

# At least one provider key is required
GEMINI_API_KEY=AIza...
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
DEEPSEEK_API_KEY=sk-...
XAI_API_KEY=xai-...

# Active model (also changeable via Web UI)
AION_MODEL=gemini-2.5-flash

# Messaging (optional)
TELEGRAM_BOT_TOKEN=123456:ABC...
TELEGRAM_CHAT_ID=123456789
DISCORD_BOT_TOKEN=...
SLACK_BOT_TOKEN=xoxb-...
SLACK_APP_TOKEN=xapp-...

# Web server (optional)
AION_PORT=7000
AION_HOST=127.0.0.1   # use 0.0.0.0 for LAN access

# Auto-updates (optional)
AION_GITHUB_REPO=xynstr/aion

config.json

Managed via Web UI (System tab) or aion config set <key> <value>.

Key Default Description
model gemini-2.5-flash Active LLM model
check_model (unset) Model for internal YES/NO completion checks
max_history_turns 40 Conversation history limit
tts_engine edge TTS engine: edge, sapi5, pyttsx3, off
tts_voice (provider default) TTS voice name
thinking_level standard Reasoning depth: off / minimal / standard / deep / extreme
browser_headless true Playwright headless mode
tool_tier 1 Tool schema verbosity: 1 = core only, 2 = all
web_auth_token (unset) Bearer token for Web UI (set by web_tunnel plugin)

Task Routing

Route tasks to the best model automatically:

"task_routing": {
  "coding":   "claude-opus-4-6",
  "review":   "claude-sonnet-4-6",
  "browsing": "gemini-2.5-flash",
  "default":  "gemini-2.5-pro"
}

Configure via Web UI (System → Task Routing) or in chat:

"Use Claude for coding, Gemini for everything else"

File Structure

aion/
├── aion.py                      # Core: memory, LLM loop, tool dispatch
├── aion_web.py                  # Web server (FastAPI + SSE)
├── aion_cli.py                  # CLI mode
├── aion_launcher.py             # Entry point (aion command)
├── plugin_loader.py             # Plugin loading + hot-reload
├── config_store.py              # Thread-safe config I/O
├── static/index.html            # Web UI (Vanilla JS, no build step)
├── plugins/                     # All plugins (one folder per plugin)
├── prompts/rules.md             # System prompt / behavioral rules
├── character.md                 # AION's personality (evolves automatically)
├── AION_SELF.md                 # Technical self-documentation
├── aion_memory.json             # Persistent memory (max 300 entries)
├── conversation_history.jsonl   # Full conversation history
├── config.json                  # Active settings
└── .env                         # API keys (git-ignored)

Troubleshooting

Problem Solution
Menu shows raw ANSI codes Run aion update — VT100 is auto-enabled on Windows
aion --setup does nothing pip install -r requirements.txt && pip install -e . first
No module named 'google' pip install google-genai
Plugin not loading Check aion_events.log — usually a missing pip install
Port already in use Set AION_PORT=7001 in .env
Playwright not found python -m playwright install chromium
ffmpeg not in PATH Add ffmpeg to PATH or install via winget
Voice messages not working Install ffmpeg + set tts_engine=edge in config
Discord/Slack bot silent Check token + intents/permissions in developer portal

Log files: aion_events.log, aion_start.log