AI Agent
The AI agent is the core of Aether. It runs inside each workspace and can write code, execute commands, navigate a browser, provision infrastructure, and manage databases — all from natural language prompts.
Model Configuration (BYOK)
Aether uses a Bring Your Own Key model. You connect your own AI provider credentials, and Aether routes requests accordingly.
Supported Providers
| Provider | Model | Notes |
|---|
| Anthropic | Claude (via Agent SDK) | Recommended for general use |
| OpenAI | Codex | Alternative provider |
| OpenRouter | Any supported model | Access to hundreds of models |
| Codebuff | Custom routing | Routes through Codebuff backend |
Adding API Keys
Go to Settings > API Keys and add your provider key. The key is encrypted and stored securely.
aether apikey add --provider anthropic --key sk-ant-...
anthropic, openai, openrouterYou need at least one API key configured before the agent can process tasks. The agent uses the key matching the provider configured for your account.
| Tool | Description |
|---|
| File read | Read file contents by path |
| File write | Create or overwrite files |
| File edit | Make targeted edits to existing files |
| Bash execution | Run shell commands in the workspace |
| Glob | Find files by pattern |
| Grep | Search file contents with regex |
| Web search | Search the web for documentation or references |
| Web fetch | Retrieve content from a URL |
| Todo tracking | Manage task checklists during work |
| Tool | Description |
|---|
browser_navigate | Navigate to a URL |
browser_screenshot | Capture a screenshot of the current page |
browser_snapshot | Take an accessibility snapshot (structured DOM) |
browser_click | Click an element on the page |
browser_type | Type text into an input field |
browser_fill | Fill a form field directly |
browser_wait_for | Wait for an element, text, or condition |
browser_get_text | Extract text content from an element |
browser_evaluate | Execute JavaScript on the page |
browser_press_key | Press a keyboard key |
browser_go_back | Navigate back in browser history |
browser_close | Close the browser |
The browser runs inside the workspace VM, so it can access localhost URLs from your dev server. This is how the agent tests UI changes in real-time.
| Tool | Description |
|---|
provision_infra | Create a new infrastructure resource (database, cache, etc.) |
list_infra | List all provisioned resources |
get_infra_types | Get available infrastructure types |
infra_status | Check the status of a resource |
infra_logs | View logs for a resource |
infra_wait | Wait for a resource to reach a target state |
| Tool | Description |
|---|
db_query | Execute a read-only SQL query |
db_execute | Execute a write SQL statement |
db_schema | Inspect database schema (tables, columns, types) |
db_migrate | Run database migrations |
How the Agent Runs
- You create a task with a prompt (via web chat or CLI)
- The workspace service starts the agent runtime
- The agent streams responses over WebSocket — text, tool calls, thinking steps
- Tool invocations execute inside the workspace (file operations, bash commands, browser actions)
- Conversation history is persisted to the database for context continuity
- The agent continues until the task is complete, it needs your input, or you abort
The agent executes real commands in your workspace. It can install packages, modify files, start/stop processes, and provision infrastructure. Review the agent’s actions, especially for destructive operations.
