Display current server status.

Spawn a new agent from *config*. Returns the agent_id.

Stop an agent. Returns True if it existed.

Switch to a provider/model combination and return the result. Args: provider: Provider to switch to (optional)

Convert to dict for backwards compatibility using enums.

Update reasoning content during streaming. Args: reasoning: Reasoning/thinking content from model

Validate transport type.

Initialize command error. Args: message: Error message command: Command name that failed (optional)

Build server-to-tools grouping for the system prompt.

Clear the current status display without resetting state (sync version). This is used when transitioning between display modes (e.g., streami

Clear previous status lines from terminal. Assumes cursor is at the start of the first line.

Ensure every tool_call_id in the assistant message has a matching result. This is a safety net that prevents OpenAI 400 errors caused by orph

Start the interactive-shell UI. We pass in the *StreamManager* plus the ToolManager for parity with production code.

Generate system prompt for dynamic tools mode. In dynamic tools mode, the LLM has access to a tool discovery system instead of individual too

Get integer tool config from CLI/env (type-safe!).

Fallback for non-streaming clients.

Check if a registry key is a real plan ID (not a checkpoint/state).

List available tools.

Create a mock AgentManager with the given agent configs.

Build a minimal mock context for ChatUIManager.

Create a mock tool state for soft block repair tests.

Get all known plan IDs from both in-memory registry and disk. Excludes checkpoint/state files. Combines both sources because in-memor

Process a single streaming chunk. Args: raw_chunk: Raw chunk from LLM provider

Render current streaming status with debounced reasoning preview.

Resolve a full or prefix plan ID to the full UUID. Supports prefix matching: if the given ID is a prefix of exactly one plan's UUID,

Serialise *obj* to JSON, falling back to _to_serializable on error.

Setup signal handlers for clean shutdown.

Start background refresh loop.

Stop background refresh loop.

Build a mock LLM response with text only (no tool call). Uses the chuk_llm native format (top-level response key).

Add a tool response message to conversation history.

Add a trusted domain. Args: domain: Domain to trust (e.g. 'example.com')

Create an assistant message.

Send a JSON message to all connected WebSocket clients.

Cleanup UI manager resources.

Clear the client cache.

Delete a named user layout. Returns True if it existed.

Add a runtime provider, discovering models from the API if needed. Args: name: Provider name api_base: API b

Return all pending attachments and clear the staging area.

If *tool_obj* exposes a synchronous ._execute / .run / .execute / .__call__, replace that method with an async wrapper so the executor can aw

Format the command output for a specific mode. Can be overridden for mode-specific formatting.

Create ToolCallResult from chuk's ToolResult. Args: tool_result: chuk_tool_processor.models.tool_result.ToolResult

Get tools adapted for LLM with name mapping.

Return config and context for an agent (used by group_store). Returns ``{config: AgentConfig, context: ChatContext}`` or None.

Return status dict for a single agent, or None.

Parse arguments JSON string to dict.

Get default config file path.

Get conversation length (excluding system prompt).

Get the currently active server.

Get disabled tools by specific reason - accepts enum or string!

Get path to chat history file.

Get a human-readable status indicator for a provider's token. Args: provider_name: Provider name token_manager: TokenManager

Convenience function to create a RuntimeConfig instance. Args: mcp_config: The loaded MCPConfig (defaults to getting from ConfigMana

Get server for tool from stream_manager.

Get list of available tool names for plan validation. Returns: List of tool name strings.

Launch chat mode for testing with stream_manager. Separated from main function to keep it clean. Args: stream_manager: Test str

Handle the /interrupt command with streaming awareness. Args: ui: UI manager instance Returns: True if command was hand

Check if a provider was added at runtime.

Check if environment variable is set. Args: var: EnvVar enum member Returns: True if variable is set (even if empty string)

Check if a tool is enabled (not disabled).

Check if a tool is enabled.

Async load from file (future-proof for async I/O).

Load execution checkpoint for resume.

Create progress callbacks that print step progress.

Notify dashboard that an MCP App closed.

Return pending attachments without clearing.

Print tool call notification before execution. Note: This is called but the output is immediately cleared by streaming display. The a

Build an image_url content block from a URL (no download).

Publish an artifact to the shared store.

Push tool input to the app (sent after initialization).

Read a resource by URI. Args: uri: Resource URI to read (e.g. ui://tool-demo/chart) server_name: Optional server name

Regenerate system prompt with current tools.

Reload configuration from file.

Restore original signal handlers.

Revalidate all tools and return summary.

Execute a plan and return (result, plan_start_time).

Save all agents' configs and sessions to disk. Parameters ---------- agent_manager: The AgentManager whose agents to save. de

Send a message from one agent to another. The message is injected into the target agent's input queue as a system-annotated prompt.