Get timeout by enum (type-safe).

Get a command by name. Args: name: Command name (can include subcommand like 'tools list'). mode: Optional m

Register a command. Args: command: The command to register. group: Deprecated, ignored. Groups register as r

Create a successful validation result.

Run a command and show the result.

Process the conversation loop, handling tool calls and responses with streaming. Args: max_turns: Maximum number of conversation

Handle slash command. Args: user_input: User input string Returns: True if handled as command, False otherwi

Build a minimal mock chat_context.

Factory method for convenient creation. Args: tool_manager: Tool management interface provider: Provider to

Run the demo application.

Process a new chunk and update state.

Create a mock ModelManager with standard methods.

Execute tool_calls in parallel using ToolManager.stream_execute_tools(). Args: tool_calls: List of tool call objects fro

Extract the actual value from MCP response structures. MCP responses can be nested in various ways: 1. Direct value (number, string)

Convert Rich color names to prompt_toolkit format. Args: color_str: Rich color string (e.g., "bright_white", "dim", "bold yellow")

Sync helper — returns a ChatContext that has had _system_prompt set.

Load session data from disk. If the file isn't found in the agent-namespaced directory, checks the flat root directory for backward c

Initialize the session with system prompt and context management.

Execute a plan with parallel batch execution. Steps are grouped into topological batches. Steps within a batch have no dependencies o

Create from dict.

Register a bridge for the given agent_id.

Store or update a memory entry (upsert by key).

Return a mock DashboardBridge.

Return a mock DashboardServer.

Clear all entries in a scope. Returns count of removed entries.

Add a subcommand to this group.

Execute tool and return ToolCallResult. When middleware is enabled (default), CTP handles: - Retry with exponential backoff for trans

Start a new streaming operation.

Parse command line arguments into kwargs. Simple argument parser that handles: - Flags: --flag or -f - Options: --op

Convenience function to get the current application context. Returns: The current ApplicationContext Raises: RuntimeErr

Load and parse MCP config file with token resolution (sync). For async contexts, prefer load_async() to avoid blocking the event loop.

Check if a tool is a polling/status tool that should be exempt from loop detection. Polling tools (like render_status, check_progress, etc.)

Build a mock ChatUIManager.

Route an inbound browser message to the appropriate bridge.

Get command completions for partial input. Args: partial_text: The partially typed command. Returns:

Detect HTTP/SSE/STDIO servers from config and populate server lists.

Close StreamManager and cleanup.

Log a finding at a specific layer.

Configure centralized logging for MCP CLI and all dependencies. Args: level: Base logging level (DEBUG, INFO, WARNING, ERROR, CRITIC

Show final tool execution result. Args: tool: Tool execution state

Stop streaming and return final content. Args: interrupted: Whether streaming was interrupted by user Returns:

Dispatch a broadcast through the router or directly to the server.

Register tools found by discovery operations. Extracts tool names from search_tools, list_tools, or get_tool_schema results and regis

Get active provider.

Alias for load_sync for backward compatibility.

Update context with keyword arguments. This method provides a simple way to update multiple context attributes.

Return a (bridge, mock_server) pair.

Build a mock ChatContext.

Handle a JSON-RPC message from the browser. Returns a JSON-RPC response string, or *None* for notifications.

Build a mock StreamingDisplayManager.

Convenience function to execute a unified command from CLI. This is a simplified interface for main.py to use when migrating from legacy

Initialize by parsing config, setting up OAuth, and creating StreamManager. Args: namespace: Default namespace for tools

Save preferences to file.

Build a mock ConversationProcessor.

Get or create the global preference manager instance.

Extract a clean result value from ToolCallResult.result. MCP tool results come in several forms depending on the execution path: 1. CTP midd

Try to parse a string as a number, return original if not possible.

Add user message to conversation. Accepts plain text (str) or multi-modal content blocks (list[dict]). Multi-modal content is injecte

Get all tools, converted to mcp-cli ToolInfo model.

Load a plan by ID (or unique prefix), returned as a dict. Reads from the saved JSON file directly to preserve depends_on fields that

Add tool result to conversation history.

Generate a concise system prompt for the assistant. Note: Tool definitions are passed via the API's tools parameter, so we don't duplicate th

Set a value in context (for compatibility with dict-based code).

Add a custom OpenAI-compatible provider. API keys should be set via environment variables. Default env var pattern: {PROVIDER_NAME}_A

Get information about connected servers.

Get timeout with priority resolution (sync).

Get a tool by name, optionally filtering by namespace. Uses O(1) index lookup instead of scanning all tools.

Handle a slash command. Returns True if should continue.

Check if a server is disabled in preferences. Args: server_name: Name of the server to check Returns: True i

Process tool calls from a chunk and accumulate them. Args: chunk_tool_calls: Tool calls extracted from current chunk

Read a local file and build content blocks. Raises ------ FileNotFoundError If the file does not exist. ValueError If

Process CLI options. Sets up environment, triggers discovery, and parses config. ENHANCED: Now validates server configuration and provides be

Build a UniversalPlan from a dict and save it. Saves the plan with dependency information preserved. The upstream PlanRegistry's to_d

Set a custom ToolManager factory for testing. Args: factory: A callable that creates ToolManager instances, or None to reset.

Execute multiple tool calls in parallel, yielding results as they complete. This is the streaming version of execute_tools_parallel - result

Test a single command and report results.

Build tool result content, using multi-block for multimodal pages. Returns a JSON string for text/structured, or a list of content blocks

Build a standard mock tool state.

Run the main chat loop with enhanced streaming support. Args: ui: UI manager with streaming coordination ctx: Chat context

Convert a value to a JSON-safe representation.

Start a websockets server for this app.

Track transport failures for recovery detection.