> ## Documentation Index
> Fetch the complete documentation index at: https://v2.spacedrive.com/llms.txt
> Use this file to discover all available pages before exploring further.

# API

<Note type="warning">
  The API infrastructure is under active development. Some features described here are planned but not yet fully implemented. This document reflects both current implementation and planned architecture.
</Note>

The API infrastructure provides a secure, unified entry point for all Spacedrive operations. It handles authentication, authorization, session management, and request routing across CLI, Tauri desktop app, and native applications.

## Architecture Overview

All API requests flow through a common pipeline:

```
Application → SessionContext → ApiDispatcher → PermissionLayer → Operation
```

This design enforces security policies consistently while providing rich context to operations.

## Session Management

Sessions encapsulate all request context, replacing simple parameter passing with structured metadata:

```rust theme={null}
// Create a device session (current implementation)
let session = SessionContext::device_session(device_uuid, device_name)
    .with_library(library_uuid);

// Session includes pre-configured permissions and metadata
```

Sessions include:

**Authentication Info**: Device or user credentials and authorization level.

**Library Context**: Currently selected library for scoped operations.

**Permission Set**: Granular permissions for the session.

**Request Metadata**: Tracking information for auditing and debugging.

## API Dispatcher

The dispatcher provides a unified interface for all operations:

```rust theme={null}
// Execute a library action
let result = api.execute_library_action(
    CreateLocationAction { path: "/Users/data" },
    session
).await?;

// Execute a core query
let libraries = api.execute_core_query(
    ListLibrariesQuery {},
    session
).await?;
```

The dispatcher automatically:

* Validates session permissions
* Ensures library context for library operations
* Tracks request metadata
* Converts errors to API errors

## Permission System

Permissions control access to specific operations:

```rust theme={null}
#[derive(Debug, Clone)]
pub struct PermissionSet {
    pub core: CorePermissions,
    pub library: LibraryPermissions,
    pub network: NetworkPermissions,
    pub jobs: JobPermissions,
}
```

Each category has granular permissions:

```rust theme={null}
pub struct LibraryPermissions {
    pub read_entries: bool,
    pub write_entries: bool,
    pub manage_locations: bool,
    pub manage_tags: bool,
    pub execute_jobs: bool,
}
```

### Permission Enforcement

Configure enforcement behavior:

```rust theme={null}
// Development: Permissive mode (allows all operations)
let api = ApiDispatcher::permissive(core_context);

// Production: Enforce permissions
let api = ApiDispatcher::new(core_context);
```

### Authorization Levels

Four hierarchical levels provide increasing access:

**None**: No authentication required (public operations only).

**Device**: Authenticated device with basic permissions.

**User**: Authenticated user with full library access.

**Admin**: System administrator with all permissions.

Higher levels inherit permissions from lower levels.

## Error Handling

API errors provide structured information for clients:

```rust theme={null}
pub enum ApiError {
    // Authentication/authorization failures
    Unauthorized { required_level: AuthLevel },
    PermissionDenied { operation: String, permission: String },

    // Validation errors
    ValidationError { field: String, message: String },
    InvalidInput { details: String },

    // Execution errors
    OperationFailed { operation: String, error: String },
    ResourceNotFound { resource: String, id: String },

    // System errors
    DatabaseError(String),
    InternalError(String),
}
```

Errors automatically map to HTTP status codes for REST endpoints.

## Middleware Pipeline

<Note type="warning">
  Middleware pipeline is partially implemented. The infrastructure exists but middleware chaining is not yet active.
</Note>

Cross-cutting concerns are handled through composable middleware:

```rust theme={null}
let pipeline = MiddlewarePipeline::new()
    .with_logging()
    .with_metrics()
    .with_rate_limiting();
```

### Logging Middleware

Tracks all requests with timing information:

```
[API] files.copy started (request_id: 7f3a2b1c)
[API] files.copy completed in 145ms (request_id: 7f3a2b1c)
```

### Metrics Middleware

Records operation metrics for monitoring:

* Request count by operation
* Response times (p50, p90, p99)
* Error rates by type
* Active request count

## Request Sources

Track where API calls originate:

```rust theme={null}
pub enum RequestSource {
    Cli,          // Command-line interface
    Swift,        // iOS/macOS native application
    Internal,     // Internal system calls
    Other(String), // Other sources (including Tauri desktop app)
}
```

<Note>
  Tauri desktop app currently uses `Other("tauri")` or `Internal`. A dedicated `Tauri` variant is planned.
</Note>

This enables source-specific behavior like CLI-friendly error messages.

## API Discovery

<Note type="warning">
  API discovery is a planned feature, not yet implemented.
</Note>

Applications will be able to query available operations:

```rust theme={null}
// Planned feature
let surface = api.describe_api();

for op in surface.operations {
    println!("{} - {}", op.name, op.description);
    println!("  Permissions: {:?}", op.required_permissions);
    println!("  Auth Level: {:?}", op.required_auth_level);
}
```

This will power dynamic UI generation and permission checking.

## Integration Examples

### CLI Integration

```rust theme={null}
// Create session from CLI context
let session = api.create_base_session()?
    .with_library(current_library_id());

// Execute operation
match api.execute_library_action(action, session).await {
    Ok(result) => println!("Success: {:?}", result),
    Err(ApiError::InsufficientPermissions { reason }) => {
        eprintln!("Permission denied: {}", reason);
    }
    Err(e) => eprintln!("Error: {}", e),
}
```

### Tauri Integration

The Tauri desktop app integrates with the core through Rust FFI:

```rust theme={null}
// Tauri command handlers call the API dispatcher
#[tauri::command]
async fn create_location(
    path: String,
    state: tauri::State<'_, AppState>,
) -> Result<Location, String> {
    let session = state.api.create_base_session()
        .map_err(|e| e.to_string())?
        .with_library(state.current_library_id);

    state.api.execute_library_action(
        CreateLocationAction { path },
        session
    ).await.map_err(|e| e.to_string())
}
```

The Tauri app can run the daemon in two modes:

* **In-Process**: Daemon runs within the Tauri app process
* **Background Process**: Daemon runs as a separate system process (requires macOS entitlements)

See [apps/tauri/DAEMON\_SETUP.md](../../apps/tauri/DAEMON_SETUP.md) for configuration details.

### Swift Integration

```swift theme={null}
// Swift SDK uses generated types with session management
let session = SessionContext(
    auth: deviceAuth,
    library: currentLibrary.id,
    source: .swift
)

let location = try await api.libraries.createLocation(
    CreateLocationInput(path: "/Users/data"),
    session: session
)
```

## Security Best Practices

**Always use strict permissions in production**: Development warnings help identify missing permissions during development.

**Validate library context**: Library operations must have `current_library_id` set in the session.

**Track request metadata**: Use request IDs for debugging distributed operations.

**Handle errors appropriately**: Convert internal errors to user-friendly messages at the boundary.

## Future Enhancements

Planned improvements include:

**User Authentication**: OAuth and session token support.

**Rate Limiting**: Per-user and per-operation limits.

**Request Caching**: Smart caching for read operations.

**WebSocket Support**: Real-time operation subscriptions.

<Note>
  The API infrastructure is designed to be transport-agnostic. While examples
  show Rust usage, the same patterns apply to REST endpoints and FFI boundaries.
</Note>