Configuration

Environment Variables

Abbado is configured through environment variables. There is no configuration file.

Variable	Default	Description
`IA_IDE_DATA_DIR`	`./data`	Directory for SQLite database, worktrees, and run state
`IA_IDE_BIND`	`127.0.0.1:3000`	Address and port the server binds to
`IA_IDE_FRONTEND_DIR`	`./frontend/dist`	Path to the built frontend (production mode only)
`RUST_LOG`	`abbado=debug`	Log level using the `tracing` crate filter syntax

Examples

Run on a different port:

IA_IDE_BIND=0.0.0.0:8080 cargo run

Custom data directory:

IA_IDE_DATA_DIR=/opt/abbado/data cargo run

Verbose logging:

RUST_LOG=abbado=trace cargo run

Quiet logging (errors only):

RUST_LOG=abbado=error cargo run

Data Directory

The data directory (IA_IDE_DATA_DIR) contains all persistent state:

$IA_IDE_DATA_DIR/
├── abbado.db             # SQLite database
└── runs/
    └── <run-uuid>/
        ├── worktree/
        │   └── <repo-name>/    # Git worktree
        ├── cli-state/          # CLI session state (for resume)
        ├── hook.sh             # Hook script for this run
        └── claude-hooks.json   # Hook configuration

The data directory is created automatically on first run. It is gitignored in the Abbado repository.

SQLite Database

Abbado uses a single SQLite database (abbado.db) for all persistent data:

projects — project records
project_repos — repos linked to projects
agents — agent configurations
runs — run metadata and status
events — immutable event log
settings — key-value configuration store

Migrations run automatically on startup. The database is created if it doesn’t exist.

Run Directories

Each run gets its own directory under $IA_IDE_DATA_DIR/runs/<run-uuid>/. This includes:

worktree/ — git worktrees for each repo (where the agent works)
cli-state/ — CLI session state for pause/resume
hook.sh — the hook script that posts events to Abbado
claude-hooks.json — hook configuration passed to the CLI via --settings

Deleting a run via the API cleans up all of these.

Settings API

Runtime configuration is stored in the SQLite settings table as key-value pairs. These are managed through the API:

Provider Models

GET /api/settings/models
PUT /api/settings/models

Controls which providers and models appear in the agent creation form.

Schema:

{
  "providers": [
    {
      "id": "claude-code",
      "name": "Claude",
      "models": [
        {
          "id": "claude-sonnet-4-6",
          "name": "Sonnet 4.6",
          "description": "Fast, balanced"
        }
      ]
    }
  ]
}

Defaults:

Provider	Models
Claude (`claude-code`)	Sonnet 4.6, Opus 4.6, Haiku 4.5
ChatGPT (`codex`)	o3, o4-mini, GPT-4.1

Notification Settings

GET /api/settings/notifications
PUT /api/settings/notifications

Schema:

{
  "discord_webhook_url": "https://discord.com/api/webhooks/...",
  "notify_on_pipeline_complete": true,
  "notify_on_run_failed": true
}

Field	Type	Default	Description
`discord_webhook_url`	string	`""`	Discord webhook URL. Empty disables notifications.
`notify_on_pipeline_complete`	bool	`true`	Notify when sessions complete or go idle
`notify_on_run_failed`	bool	`true`	Notify when runs fail

Frontend Configuration

The frontend dev server (Vite) proxies API requests to http://localhost:3000. This is configured in the Vite config and requires no manual setup.

For production, build the frontend with npm run build and set IA_IDE_FRONTEND_DIR to point to the dist directory. The Rust server serves the static files directly.

cd frontend && npm run build
IA_IDE_FRONTEND_DIR=./frontend/dist cargo run --release

Network

Abbado binds to 127.0.0.1:3000 by default — localhost only. To expose it on the network:

IA_IDE_BIND=0.0.0.0:3000 cargo run

The server uses:

HTTP for REST API and SSE
WebSocket for terminal connections (upgrades from HTTP)
No TLS — use a reverse proxy (nginx, Caddy) if you need HTTPS