CLI Reference
Author
Usage
aether [flags]
Aether opens an interactive session. No subcommands needed — everything happens through slash commands once it's running.
Flags
| Flag | Description |
|---|---|
--no-animation | Skip startup animation |
--version, -v | Show version |
Interactive Commands
Once inside the session, use slash commands (case-insensitive: /CONFIG, /Genesis, etc. all work):
| Command | Description |
|---|---|
/help | Show available commands |
/config | Configure your AI provider and model |
/genesis | Analyze the project and generate documentation |
/sync | Update existing docs incrementally |
/prompt | Turn a task into an optimized prompt for another AI |
/ask | Ask a question, answered from your project docs |
/html | Build a browsable single-file HTML viewer for the docs (free — no tokens) |
/exclude | Manage paths excluded from scans |
/clean | Manage global data (caches, configs, projects) |
/cleancode | Scan the project for clean-code violations and write a report |
/clear | Clear the screen |
/exit | Exit Aether |
Every command also accepts help, --help, or -h for detailed usage — e.g. /config help.
Error Handling & Retries
All API calls retry automatically with exponential backoff:
| Error type | Retries | Backoff | Behavior |
|---|---|---|---|
| Rate limit (429) | Up to 6 | 15s, 30s, 60s, 120s... | Respects provider's retry-after hint if present |
| Server errors (5xx) | Up to 3 | 2s, 4s, 8s | Standard exponential backoff |
| Timeouts | Up to 3 | 2s, 4s, 8s | Idle timeout (no data for 2 min), not total time |
Rate limits from free tiers (Gemini's 15 RPM, for example) are handled automatically — the CLI waits and retries instead of failing. During generation, each step shows its retry status inline.
Dropdown Navigation
Type / to see all commands. Use ↑↓ arrows to navigate, Enter to select, Tab to autocomplete.
Planned
- MCP Server — Expose knowledge base to Claude/Cursor/Gemini
Command Pages
/config— Configure your AI provider and model/genesis— Analyze the project and generate documentation/sync— Update existing docs incrementally/prompt— Turn a task into an optimized prompt for another AI/ask— Ask a question, answered from your project docs/html— Build a browsable single-file HTML viewer for the docs/exclude— Manage paths excluded from scans/clean— Manage global data (caches, configs, projects)/cleancode— Scan the project for clean-code violations and write a report