arcade-mcp/libs/arcade-cli/arcade_cli/main.py
Eric Gustin 113d0d3086
CLI Usage (#593)
TLDR; 

The philosophy of CLI usage is "fire and forget" and "best effort". You
can opt out by setting `ARCADE_USAGE_TRACKING=0`.

We are capturing two events: `CLI execution succeeded` and `CLI
execution failed`. Reporting to PostHog is a short lived (maximum 10
seconds) subprocess that does not block the main CLI execution process.

`~/.arcade/usage.json` persists two values `anon_id` and
`linked_principal_id`. The logged in status of the CLI user determines
which ID is used. Upon `arcade login`, the `anon_id` is aliased with
`linked_principal_id`. Upon `arcade logout` the `linked_principal_id` is
removed and the `anon_id` is rotated.

## CLI Usage Tracking - How It Works

The usage tracking system implements an identity management and event
tracking pipeline. Here's how the pieces work together:

### **Identity State Management (`usage.json`)**

The system maintains a persistent identity file at
`~/.arcade/usage.json` with this structure:
```json
{
  "anon_id": "uuid",
  "linked_principal_id": "uuid" | null
}
```

**Key mechanics:**
- **`anon_id`**: Generated once on first CLI use and persists across
sessions. This UUID tracks all anonymous activity.
- **`linked_principal_id`**: Initially `null`. Once the user logs in and
we successfully alias their identity, this field stores their
`principal_id` to indicate this `anon_id` has been linked.
- **Atomic writes**: All updates use a temp file + atomic rename pattern
to prevent corruption from concurrent CLI processes
- **File locking**: Uses `fcntl` (Unix) to coordinate reads/writes
across multiple simultaneous CLI invocations
- **In-memory cache**: The `UsageIdentity` class caches the loaded data
to avoid repeated file I/O within a single CLI invocation

### **Identity Resolution Flow**

When tracking an event, the system determines the `distinct_id` (who to
attribute the event to) via this waterfall:

1. **Check `linked_principal_id`** in `usage.json`
   - If present → use it (user was previously aliased)
   - This is the fastest path and avoids API calls

2. **Fetch `principal_id` from Arcade Cloud API**
- Makes HTTP request to `/api/v1/auth/validate` with the user's API key
from `~/.arcade/credentials.yaml`
   - If authenticated → returns `principal_id`
   - Has 2s timeout for responsiveness

3. **Fall back to `anon_id`**
   - If not authenticated or API call fails → use anonymous ID
   - Marks event with `is_anon=True` flag

### **The Aliasing Lifecycle**

PostHog aliasing links anonymous activity to authenticated users. Here's
the state machine:

#### **Stage 1: Anonymous User**
```
usage.json: { "anon_id": "abc-123", "linked_principal_id": null }
All events → sent with distinct_id="abc-123" and is_anon=True
```

#### **Stage 2: Login Event**
1. User runs `arcade login`
2. Command completes successfully (auth token saved)
3. `CommandTracker` detects successful login
4. Fetches `principal_id` from API
5. Checks `should_alias()` → returns `True` because
`linked_principal_id` is `null`
6. **Calls `alias()` synchronously** (blocking):
   ```python
   posthog.alias(previous_id="abc-123", distinct_id="zyx-321")
   ```
7. Updates `usage.json`:
   ```json
   { "anon_id": "abc-123", "linked_principal_id": "zyx-321" }
   ```
8. PostHog backend merges all events with `distinct_id="abc-123"` into
the user profile for `"zyx-321"`

#### **Stage 3: Authenticated User**
```
usage.json: { "anon_id": "abc-123", "linked_principal_id": "zyx-321" }
All events → sent with distinct_id="zyx-321" and is_anon=False
```
- Events are directly attributed to the authenticated user
- No more API calls needed (uses cached `linked_principal_id`)

#### **Stage 4: Logout Event**
1. User runs `arcade logout`
2. Logout event is sent with the authenticated `distinct_id`
3. `CommandTracker` detects successful logout
4. **Rotates identity** by calling `reset_to_anonymous()`:
   ```json
   { "anon_id": "xyz-789", "linked_principal_id": null }
   ```
5. New `anon_id` prevents cross-contamination if another user logs in

### **Critical Constraint: Alias Timing**

PostHog requires that `alias()` is called **BEFORE** any events are sent
with the new `distinct_id`. This is why:
- **`alias()` is synchronous (blocking)**: Guarantees it completes
before the login success event is sent
- **Subsequent events use `linked_principal_id`**: Once aliased, all
future events use the authenticated ID
- **Lazy aliasing**: If a user authenticates via another mechanism (not
through `arcade login`), the system detects this on the next command and
performs aliasing before sending that command's event

### **Event Capture Pipeline**

When `CommandTracker.track_command_execution()` is called:

1. **Resolve identity** → determines `distinct_id` and `is_anon` flag
2. **Build event properties**:
   ```python
   {
     "command_name": "toolkit.run",
     "cli_version": "1.2.3",
     "python_version": "3.11.0",
     "os_type": "Darwin",
     "os_release": "23.4.0",
     "duration": 1250.42,  # milliseconds
     "error_message": "..."  # if failed
   }
   ```
3. **Call `UsageService.capture()`**:
   - Serializes event data to JSON
   - Spawns detached subprocess: `python -m arcade_cli.usage`
   - Passes data via `ARCADE_USAGE_EVENT_DATA` env var
   - **Returns immediately** (non-blocking)

4. **Detached subprocess (`__main__.py`)**:
   - Runs independently, survives parent CLI exit
   - Deserializes event data
- If `is_anon=True`, sets `$process_person_profile=False` (tells PostHog
not to create a full profile)
   - Sends event to PostHog with 5s timeout
   - Exits (hard exit after 10s max via timeout thread)

### **Concurrency Handling**

Multiple CLI processes can run simultaneously. The system handles this
via:
- **File locking** on `usage.json` (shared lock for reads, exclusive for
writes)
- **Atomic writes** via temp files ensure incomplete writes never
corrupt the file
- **Idempotent aliasing**: `should_alias()` prevents redundant alias
calls

### **Edge Cases Handled**

1. **Side-channel authentication**: User authenticates outside of
`arcade login` (e.g., manually editing credentials)
   - Detected via "lazy aliasing" check on every command
- Performs alias if `linked_principal_id` doesn't match current
`principal_id`

2. **API failures during identity fetch**: Falls back to anonymous
tracking
   - 2s timeout prevents hanging
   - Silent failure doesn't disrupt CLI

3. **PostHog merge restrictions**: Can't alias returning users who
already have a profile
- System stores `linked_principal_id` to avoid retrying impossible
aliases
   - New users (never logged in before) get full history stitched

4. **Multiple accounts on same machine**: Logout rotates `anon_id`
   - User A's anonymous activity won't leak into User B's profile

### **Privacy & Performance**

- **Opt-out**: `ARCADE_USAGE_TRACKING=0` disables all tracking
- **Non-blocking**: Events never slow down CLI (detached subprocess)
- **Anonymous profiles**: `$process_person_profile=False` for `anon_id`
events minimizes data collection
- **Silent failures**: Network issues or PostHog errors never surface to
users
2025-10-03 10:15:08 -07:00

832 lines
27 KiB
Python

import asyncio
import os
import subprocess
import sys
import threading
import uuid
import webbrowser
from pathlib import Path
from typing import Optional
import httpx
import typer
from arcadepy import Arcade
from rich.console import Console
from rich.text import Text
from tqdm import tqdm
import arcade_cli.secret as secret
import arcade_cli.worker as worker
from arcade_cli.authn import LocalAuthCallbackServer, check_existing_login
from arcade_cli.constants import (
CREDENTIALS_FILE_PATH,
PROD_CLOUD_HOST,
PROD_ENGINE_HOST,
)
from arcade_cli.deployment import Deployment
from arcade_cli.display import (
display_eval_results,
)
from arcade_cli.show import show_logic
from arcade_cli.toolkit_docs import generate_toolkit_docs
from arcade_cli.usage.command_tracker import TrackedTyper, TrackedTyperGroup
from arcade_cli.utils import (
Provider,
compute_base_url,
compute_login_url,
get_eval_files,
handle_cli_error,
load_eval_suites,
log_engine_health,
require_dependency,
resolve_provider_api_key,
validate_and_get_config,
version_callback,
)
cli = TrackedTyper(
cls=TrackedTyperGroup,
add_completion=False,
no_args_is_help=True,
pretty_exceptions_enable=True,
pretty_exceptions_show_locals=False,
pretty_exceptions_short=True,
rich_markup_mode="markdown",
context_settings={"help_option_names": ["-h", "--help"]},
help="Arcade CLI - Build, deploy, and manage MCP servers and AI tools. Create new projects, run servers with multiple transports, configure clients, and deploy to Arcade Cloud.",
epilog="Pro tip: use --help after any command to see command-specific options.",
)
cli.add_typer(
worker.app,
name="server",
help="Manage deployments of tool servers (logs, list, etc)",
rich_help_panel="Manage",
)
cli.add_typer(
secret.app,
name="secret",
help="Manage tool secrets in the cloud (set, unset, list)",
rich_help_panel="Manage",
)
console = Console()
@cli.command(help="Log in to Arcade Cloud", rich_help_panel="User")
def login(
host: str = typer.Option(
PROD_CLOUD_HOST,
"-h",
"--host",
help="The Arcade Cloud host to log in to.",
),
port: Optional[int] = typer.Option(
None,
"-p",
"--port",
help="The port of the Arcade Cloud host (if running locally).",
),
callback_host: str = typer.Option(
None,
"--callback-host",
help="The host to use to complete the auth flow - this should be the same as the host that the CLI is running on. Include the port if needed.",
),
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
) -> None:
"""
Logs the user into Arcade Cloud.
"""
if check_existing_login():
console.print("\nTo log out and delete your locally-stored credentials, use ", end="")
console.print("arcade logout", style="bold green", end="")
console.print(".\n")
return
# Start the HTTP server in a new thread
state = str(uuid.uuid4())
auth_server = LocalAuthCallbackServer(state)
server_thread = threading.Thread(target=auth_server.run_server)
server_thread.start()
try:
# Open the browser for user login
login_url = compute_login_url(host, state, port, callback_host)
console.print("Opening a browser to log you in...")
if not webbrowser.open(login_url):
console.print(
f"If a browser doesn't open automatically, copy this URL and paste it into your browser: {login_url}",
style="dim",
)
# Wait for the server thread to finish
server_thread.join()
except KeyboardInterrupt:
auth_server.shutdown_server()
except Exception as e:
handle_cli_error("Login failed", e, debug)
finally:
if server_thread.is_alive():
server_thread.join() # Ensure the server thread completes and cleans up
@cli.command(help="Log out of Arcade Cloud", rich_help_panel="User")
def logout(
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
) -> None:
"""
Logs the user out of Arcade Cloud.
"""
try:
# If the credentials file exists, delete it
if os.path.exists(CREDENTIALS_FILE_PATH):
os.remove(CREDENTIALS_FILE_PATH)
console.print("You're now logged out.", style="bold")
else:
console.print("You're not logged in.", style="bold red")
except Exception as e:
handle_cli_error("Logout failed", e, debug)
@cli.command(
help="Create a new server package directory. Example usage: `arcade new my_mcp_server`",
rich_help_panel="Build",
)
def new(
server_name: str = typer.Argument(
help="The name of the server to create",
metavar="SERVER_NAME",
),
directory: str = typer.Option(os.getcwd(), "--dir", help="tools directory path"),
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
full: bool = typer.Option(
False,
"--full",
"-f",
help="Create a starter MCP server (pyproject.toml, server.py, .env.example)",
),
) -> None:
"""
Creates a new MCP server with the given name
"""
from arcade_cli.new import create_new_toolkit, create_new_toolkit_minimal
try:
if not full:
create_new_toolkit_minimal(directory, server_name)
else:
create_new_toolkit(directory, server_name)
except Exception as e:
handle_cli_error("Failed to create new server", e, debug)
@cli.command(
name="mcp",
help="Run MCP servers with different transports",
rich_help_panel="Run",
)
def mcp(
transport: str = typer.Argument("http", help="Transport type: stdio, http"),
host: str = typer.Option("127.0.0.1", "--host", help="Host to bind to (HTTP mode only)"),
port: int = typer.Option(8000, "--port", help="Port to bind to (HTTP mode only)"),
tool_package: Optional[str] = typer.Option(
None,
"--tool-package",
"--package",
"-p",
help="Specific tool package to load (e.g., 'github' for arcade-github)",
),
discover_installed: bool = typer.Option(
False, "--discover-installed", "--all", help="Discover all installed arcade tool packages"
),
show_packages: bool = typer.Option(
False, "--show-packages", help="Show loaded packages during discovery"
),
reload: bool = typer.Option(
False, "--reload", help="Enable auto-reload on code changes (HTTP mode only)"
),
debug: bool = typer.Option(False, "--debug", help="Enable debug mode with verbose logging"),
otel_enable: bool = typer.Option(
False, "--otel-enable", help="Send logs to OpenTelemetry", show_default=True
),
env_file: Optional[str] = typer.Option(None, "--env-file", help="Path to environment file"),
name: Optional[str] = typer.Option(None, "--name", help="Server name"),
version: Optional[str] = typer.Option(None, "--version", help="Server version"),
cwd: Optional[str] = typer.Option(None, "--cwd", help="Working directory to run from"),
) -> None:
"""
Run Arcade MCP Server (passthrough to arcade_mcp_server).
This command provides a unified CLI experience by passing through
all arguments to the arcade_mcp_server module.
Examples:
arcade mcp stdio
arcade mcp http --port 8080
arcade mcp --tool-package github
arcade mcp --discover-installed --show-packages
"""
# Build the command to pass through to arcade_mcp_server
cmd = [sys.executable, "-m", "arcade_mcp_server", transport]
# Add optional arguments
cmd.extend(["--host", host])
cmd.extend(["--port", str(port)])
if debug:
cmd.append("--debug")
if otel_enable:
cmd.append("--otel-enable")
if tool_package:
cmd.extend(["--tool-package", tool_package])
if discover_installed:
cmd.append("--discover-installed")
if show_packages:
cmd.append("--show-packages")
if reload:
cmd.append("--reload")
if env_file:
cmd.extend(["--env-file", env_file])
if name:
cmd.extend(["--name", name])
if version:
cmd.extend(["--version", version])
if cwd:
cmd.extend(["--cwd", cwd])
try:
# Show what command we're running in debug mode
if debug:
console.print(f"[dim]Running: {' '.join(cmd)}[/dim]")
# Execute the command and pass through all output
result = subprocess.run(cmd, check=False)
# Exit with the same code as the subprocess
if result.returncode != 0:
handle_cli_error("Failed to run MCP server")
except KeyboardInterrupt:
console.print("\n[yellow]MCP server gracefully shutdown[/yellow]")
except FileNotFoundError:
handle_cli_error(
"arcade_mcp_server module not found. Make sure arcade-mcp-server is installed"
)
@cli.command(
help="Show the installed tools or details of a specific tool",
rich_help_panel="Build",
)
def show(
server: Optional[str] = typer.Option(
None, "-T", "--server", help="The server to show the tools of"
),
tool: Optional[str] = typer.Option(
None, "-t", "--tool", help="The specific tool to show details for"
),
host: str = typer.Option(
PROD_ENGINE_HOST,
"-h",
"--host",
help="The Arcade Engine address to show the tools/servers of.",
),
local: bool = typer.Option(
False,
"--local",
"-l",
help="Show the local environment's catalog instead of an Arcade Engine's catalog.",
),
port: Optional[int] = typer.Option(
None,
"-p",
"--port",
help="The port of the Arcade Engine.",
),
force_tls: bool = typer.Option(
False,
"--tls",
help="Whether to force TLS for the connection to the Arcade Engine. If not specified, the connection will use TLS if the engine URL uses a 'https' scheme.",
),
force_no_tls: bool = typer.Option(
False,
"--no-tls",
help="Whether to disable TLS for the connection to the Arcade Engine.",
),
full: bool = typer.Option(
False,
"--full",
"-f",
help="Show full server response structure including error, logs, and authorization fields (only applies when used with -t/--tool).",
),
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
) -> None:
"""
Show the available tools or detailed information about a specific tool.
"""
if full and not tool:
console.print(
"⚠️ The -f/--full flag only affects output when used with -t/--tool flag",
style="bold yellow",
)
show_logic(
toolkit=server,
tool=tool,
host=host,
local=local,
port=port,
force_tls=force_tls,
force_no_tls=force_no_tls,
worker=full,
debug=debug,
)
@cli.command(help="Run tool calling evaluations", rich_help_panel="Build")
def evals(
directory: str = typer.Argument(".", help="Directory containing evaluation files"),
show_details: bool = typer.Option(False, "--details", "-d", help="Show detailed results"),
max_concurrent: int = typer.Option(
1,
"--max-concurrent",
"-c",
help="Maximum number of concurrent evaluations (default: 1)",
),
models: str = typer.Option(
"gpt-4o",
"--models",
"-m",
help="The models to use for evaluation (default: gpt-4o). Use commas to separate multiple models. All models must belong to the same provider.",
),
provider: Provider = typer.Option(
Provider.OPENAI,
"--provider",
"-p",
help="The provider of the models to use for evaluation.",
),
provider_api_key: str = typer.Option(
None,
"--provider-api-key",
"-k",
help="The model provider API key. If not provided, will look for the appropriate environment variable based on the provider (e.g., OPENAI_API_KEY for openai provider), first in the current environment, then in the current working directory's .env file.",
),
debug: bool = typer.Option(False, "--debug", help="Show debug information"),
) -> None:
"""
Find all files starting with 'eval_' in the given directory,
execute any functions decorated with @tool_eval, and display the results.
"""
require_dependency(
package_name="arcade_evals",
command_name="evals",
install_command=r"pip install 'arcade-mcp\[evals]'",
)
# Although Evals does not depend on the TDK, some evaluations import the
# ToolCatalog class from the TDK instead of from arcade_core, so we require
# the TDK to run the evals CLI command to avoid possible import errors.
require_dependency(
package_name="arcade_tdk",
command_name="evals",
install_command=r"pip install arcade-tdk",
)
models_list = models.split(",") # Use 'models_list' to avoid shadowing
# Resolve the API key for the provider
resolved_api_key = resolve_provider_api_key(provider, provider_api_key)
if not resolved_api_key:
provider_env_vars = {
Provider.OPENAI: "OPENAI_API_KEY",
}
env_var_name = provider_env_vars.get(provider, f"{provider.upper()}_API_KEY")
handle_cli_error(
f"API key not found for provider '{provider.value}'. "
f"Please provide it via --provider-api-key,-k argument, set the {env_var_name} environment variable, "
f"or add it to a .env file in the current directory.",
should_exit=True,
)
eval_files = get_eval_files(directory)
if not eval_files:
return
console.print("\nRunning evaluations", style="bold")
# Use the new function to load eval suites
eval_suites = load_eval_suites(eval_files)
if not eval_suites:
console.print("No evaluation suites to run.", style="bold yellow")
return
if show_details:
suite_label = "suite" if len(eval_suites) == 1 else "suites"
console.print(
f"\nFound {len(eval_suites)} {suite_label} in the evaluation files.",
style="bold",
)
async def run_evaluations() -> None:
all_evaluations = []
tasks = []
for suite_func in eval_suites:
console.print(
Text.assemble(
("Running evaluations in ", "bold"),
(suite_func.__name__, "bold blue"),
)
)
for model in models_list:
task = asyncio.create_task(
suite_func(
provider_api_key=resolved_api_key,
model=model,
max_concurrency=max_concurrent,
)
)
tasks.append(task)
# Track progress and results as suite functions complete
with tqdm(total=len(tasks), desc="Evaluations Progress") as pbar:
results = []
for f in asyncio.as_completed(tasks):
results.append(await f)
pbar.update(1)
# TODO error handling on each eval
all_evaluations.extend(results)
display_eval_results(all_evaluations, show_details=show_details)
try:
asyncio.run(run_evaluations())
except Exception as e:
handle_cli_error("Failed to run evaluations", e, debug)
@cli.command(help="Configure MCP clients to connect to your server", rich_help_panel="Manage")
def configure(
client: str = typer.Argument(
...,
help="The MCP client to configure (claude, cursor, vscode)",
),
server_name: Optional[str] = typer.Option(
None,
"--server",
"-s",
help="Name of the server to connect to (defaults to current directory name)",
),
from_local: bool = typer.Option(
False,
"--from-local",
help="Connect to a local MCP server",
is_flag=True,
),
from_arcade: bool = typer.Option(
False,
"--from-arcade",
help="Connect to an Arcade Cloud MCP server",
is_flag=True,
),
port: int = typer.Option(
8000,
"--port",
"-p",
help="Port for local servers",
),
path: Optional[Path] = typer.Option(
None,
"--path",
"-f",
exists=False,
help="Optional path to a specific MCP client config file (overrides default path)",
),
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
) -> None:
"""
Configure MCP clients to connect to your server.
Examples:
arcade configure claude --from-local
arcade configure cursor --from-local --port 8080
arcade configure vscode --from-local --path .vscode/mcp.json
arcade configure claude --from-arcade --server my_server_name
"""
from arcade_cli.configure import configure_client
try:
configure_client(
client=client,
server_name=server_name,
from_local=from_local,
from_arcade=from_arcade,
port=port,
path=path,
)
except Exception as e:
handle_cli_error(f"Failed to configure {client}", e, debug)
@cli.command(help="Deploy servers to Arcade Cloud", rich_help_panel="Run", hidden=True)
def deploy(
deployment_file: str = typer.Option(
"worker.toml",
"--deployment-file",
"-d",
help="The deployment file to deploy.",
),
cloud_host: str = typer.Option(
PROD_CLOUD_HOST,
"--cloud-host",
"-c",
help="The Arcade Cloud host to deploy to.",
hidden=True,
),
cloud_port: Optional[int] = typer.Option(
None,
"--cloud-port",
"-cp",
help="The port of the Arcade Cloud host.",
hidden=True,
),
host: str = typer.Option(
PROD_ENGINE_HOST,
"--host",
"-h",
help="The Arcade Engine host to register the server to.",
),
port: Optional[int] = typer.Option(
None,
"--port",
"-p",
help="The port of the Arcade Engine host.",
),
force_tls: bool = typer.Option(
False,
"--tls",
help="Whether to force TLS for the connection to the Arcade Engine. If not specified, the connection will use TLS if the engine URL uses a 'https' scheme.",
),
force_no_tls: bool = typer.Option(
False,
"--no-tls",
help="Whether to disable TLS for the connection to the Arcade Engine.",
),
debug: bool = typer.Option(False, "--debug", help="Show debug information"),
) -> None:
"""
Deploy a server to Arcade Cloud.
"""
config = validate_and_get_config()
engine_url = compute_base_url(force_tls, force_no_tls, host, port)
engine_client = Arcade(api_key=config.api.key, base_url=engine_url)
cloud_url = compute_base_url(force_tls, force_no_tls, cloud_host, cloud_port)
cloud_client = httpx.Client(
base_url=cloud_url, headers={"Authorization": f"Bearer {config.api.key}"}
)
# Fetch deployment configuration
try:
deployment = Deployment.from_toml(Path(deployment_file))
except Exception as e:
handle_cli_error("Failed to parse deployment file", e, debug)
with console.status(f"Deploying {len(deployment.worker)} servers"):
for worker in deployment.worker:
console.log(f"Deploying '{worker.config.id}...'", style="dim")
try:
# Discover and upload secrets
required_secret_keys = worker.get_required_secrets()
for secret_key in required_secret_keys:
secret_value = os.getenv(secret_key)
if not secret_value:
console.log(
f"⚠️ Secret '{secret_key}' not found in environment, skipping.",
style="yellow",
)
continue
try:
secret._upsert_secret_to_engine(
engine_url, config.api.key, secret_key, secret_value
)
except Exception as e:
handle_cli_error(
f"Failed to upload secret '{secret_key}'", e, debug, should_exit=False
)
else:
console.log(
f"✅ Secret '{secret_key}' uploaded successfully",
style="dim green",
)
# Attempt to deploy worker
worker.request().execute(cloud_client, engine_client)
console.log(
f"✅ Server '{worker.config.id}' deployed successfully.",
style="dim",
)
except Exception as e:
handle_cli_error(f"Failed to deploy server '{worker.config.id}'", e, debug)
@cli.command(help="Open the Arcade Dashboard in a web browser", rich_help_panel="User")
def dashboard(
host: str = typer.Option(
PROD_ENGINE_HOST,
"-h",
"--host",
help="The Arcade Engine host that serves the dashboard.",
),
port: Optional[int] = typer.Option(
None,
"-p",
"--port",
help="The port of the Arcade Engine.",
),
local: bool = typer.Option(
False,
"--local",
"-l",
help="Open the local dashboard instead of the default remote dashboard.",
),
force_tls: bool = typer.Option(
False,
"--tls",
help="Whether to force TLS for the connection to the Arcade Engine.",
),
force_no_tls: bool = typer.Option(
False,
"--no-tls",
help="Whether to disable TLS for the connection to the Arcade Engine.",
),
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
) -> None:
"""Opens the Arcade Dashboard in a web browser.
The Dashboard is a web-based Arcade user interface that is served by the Arcade Engine.
"""
try:
if local:
host = "localhost"
# Construct base URL (for both health check and dashboard)
base_url = compute_base_url(force_tls, force_no_tls, host, port)
dashboard_url = f"{base_url}/dashboard"
# Try to hit /health endpoint on engine and warn if it is down
config = validate_and_get_config()
with Arcade(api_key=config.api.key, base_url=base_url) as client:
log_engine_health(client)
# Open the dashboard in a browser
console.print(f"Opening Arcade Dashboard at {dashboard_url}")
if not webbrowser.open(dashboard_url):
console.print(
f"If a browser doesn't open automatically, copy this URL and paste it into your browser: {dashboard_url}",
style="dim",
)
except Exception as e:
handle_cli_error("Failed to open dashboard", e, debug)
@cli.command(
help=(
"Generate documentation for a server. "
"Note: make sure to have the server installed in your current Python environment "
"before running this command."
),
rich_help_panel="Document",
hidden=True,
)
def docs(
server_name: str = typer.Option(
...,
"--server-name",
"-n",
help="The name of the server to generate documentation for.",
),
server_dir: str = typer.Option(
...,
"--server-dir",
"-t",
help=(
"The path to the server root directory (where the server code is implemented). "
"Works with relative and absolute paths."
),
),
docs_dir: str = typer.Option(
...,
"--docs-dir",
"-r",
help="The path to the root of the Arcade docs repository. Works with relative and absolute paths.",
),
docs_section: str = typer.Option(
"",
"--docs-section",
"-s",
help=(
"The section of the docs to generate documentation for. E.g. 'productivity', 'sales'. "
"This should be the name of the folder in /pages/tools. "
"Defaults to an empty string (generate the docs in the root of /pages/tools)"
),
),
openai_model: str = typer.Option(
"gpt-5-mini",
"--openai-model",
"-m",
help=(
"A few parts of the documentation are generated using OpenAI API. "
"Choose one of the 'gpt-4o' and 'gpt-5' series models."
),
show_default=True,
),
openai_api_key: str = typer.Option(
None,
"--openai-api-key",
"-o",
help="The OpenAI API key. If not provided, will get it from the `OPENAI_API_KEY` env var.",
),
skip_tool_call_examples: bool = typer.Option(
False,
"--skip-tool-call-examples",
"-se",
help="Whether to skip generating tool call examples in Python and Javascript.",
show_default=True,
),
debug: bool = typer.Option(False, "--debug", "-d", help="Show debug information"),
) -> None:
if not openai_model.startswith("gpt-4o") and not openai_model.startswith("gpt-5"):
console.print(
f"Attention: '{openai_model}' is not a valid OpenAI model. "
"Please choose one of the 'gpt-4o' and 'gpt-5' series models.",
style="bold red",
)
handle_cli_error(
f"Attention: '{openai_model}' is not a valid OpenAI model. "
"Please choose one of the 'gpt-4o' and 'gpt-5' series models."
)
try:
success = generate_toolkit_docs(
console=console,
toolkit_name=server_name,
toolkit_dir=server_dir,
docs_dir=docs_dir,
docs_section=docs_section,
openai_model=openai_model,
openai_api_key=openai_api_key,
tool_call_examples=not skip_tool_call_examples,
debug=debug,
)
except Exception as error:
handle_cli_error(
message=f"Failed to generate documentation for '{server_name}' in '{docs_dir}'",
error=error,
debug=debug,
)
success = False
if success:
console.print(
f"Generated documentation for '{server_name}' in '{docs_dir}'",
style="bold green",
)
else:
console.print(
f"Failed to generate documentation for '{server_name}' in '{docs_dir}'",
style="bold red",
)
@cli.callback()
def main_callback(
ctx: typer.Context,
_: Optional[bool] = typer.Option(
None,
"-v",
"--version",
callback=version_callback,
is_eager=True,
help="Print version and exit.",
),
) -> None:
# Commands that do not require a logged in user
public_commands = {
login.__name__,
logout.__name__,
dashboard.__name__,
evals.__name__,
mcp.__name__,
new.__name__,
show.__name__,
configure.__name__,
}
if ctx.invoked_subcommand in public_commands:
return
if not check_existing_login(suppress_message=True):
handle_cli_error("Not logged in to Arcade CLI. Use `arcade login` to log in.")