Discover available models from an OpenAI-compatible API. Args: api_base: API base URL api_key: API key

Format performance metrics with color coding. Args: ping_ms: Ping time in milliseconds (None if unavailable) Returns: T

Get environment variable as list of strings. Args: var: EnvVar enum member separator: String separator (default: comma) d

Return OpenAI-format tool definitions for plan tools.

Get confirmation setting for a specific tool. Args: tool_name: Name of the tool Returns: 'always', 'never',

Inject a synthetic assistant message for conversation flow control. Use this for system-generated messages (budget exhaustion, state

Load and parse MCP config file with token resolution (async). Uses asyncio.to_thread() to avoid blocking the event loop during file I/O.

Called when the agent's status changes.

Called when a complete conversation message is emitted.

Record a tool call in session and procedural memory. Args: tool_name: Name of the tool called arguments: Arg

Remove a runtime server from preferences. Args: name: Server name to remove Returns: True if server was remo

Run a command and optionally show output.

Set environment variable (type-safe). Args: var: EnvVar enum member value: Value to set Example: >>> set_env(EnvVar.

Set and persist theme. Args: theme: Theme name to set Raises: ValueError: If theme is not valid

Type-safe timeout override.

Add an attachment to the staging area.

Convert result to conversation history string.

Extract text from a list of MCP content blocks. Content blocks can be dicts or objects with type/text attributes. Returns parsed JSON if the

Extract permissions from resource _meta.ui.permissions.

Generate system prompt from available tools (cached with dirty flag).

Get the status for a provider.

Create a mock PlanExecutionResult.

Return a mock DashboardServer whose start() resolves to the given port.

Callback when a tool starts execution.

Report progress via callback if set.

Split preview text into multiple lines at word boundaries. Args: text: Text to split (may start with '...') max_line_

Build a mock LLM response containing a tool call. Uses the chuk_llm native format (top-level tool_calls key), which is what the real client r

Add a provider at runtime (not persisted). No network I/O. For automatic model discovery, use discover_runtime_provider() instead.

Build the ``content`` field for a user message. Returns the plain string when there are no attachments or image URLs (backward compatible).

Check the status of a provider's API key. Args: provider_name: Provider name token_manager: TokenManager instance (optional)

Clear all registered commands (useful for testing).

Clear conversation history by creating a new session.

Shut down all app servers.

Mark streaming as complete.

Display a demo section header.

Extract server names from config with HTTP server support, respecting disabled status from preferences.

Parse a raw chunk from various LLM providers into normalized format. Supports multiple formats: - chuk-llm format: {'response': str,

Get the API key for a custom provider from environment variables. Args: name: Provider name Returns: API key

Get all disabled servers.

Get timeout value with priority resolution. Args: timeout_name: Name of timeout (e.g., "streaming_chunk", "tool_executio

Get number of available tools.

Get tools filtered and validated for LLM. Results are cached per provider. Cache is invalidated on tool changes (disable/enable). Dyn

Check if a provider is a custom provider.

Load runtime configuration with clean API. Args: config_path: Path to config file (default: server_config.json) cli_overrides: CL

Debug logging with rich formatting if available.

Log info message.

Extract ``@file:path`` references from message text. Returns ------- (cleaned_text, list_of_paths) *cleaned_text* has the ``@file

Retrieve memory entries by key, query, or list all.

Manually refresh models for a provider. Args: provider: Provider name (refreshes active provider if None) Retur

Refresh models for a runtime provider by querying the API. Args: config: RuntimeProviderConfig to refresh Retur

Render streaming status with optional reasoning on separate line. Args: state: Current streaming state spinner: Current spinner f

Send a tool approval request to the dashboard. The browser will show an approve/deny dialog and respond with TOOL_APPROVAL_RESPONSE.

Shut down the server and close all client connections.

Convert to immutable UnifiedServerConfig.

Update context from dictionary.

Download page content to a local file.

Extract CSP configuration from resource _meta.ui.csp.

Extract text content from MCP SDK ToolResult structure.

Get the URL for an HTTP/SSE server by name. Args: server_name: Name of the server Returns: Server URL or Non

Execute a memory scope tool (remember, recall, forget). Memory tools are persistent-memory operations that bypass the MCP ToolManager

Handle regular (non-streaming) completion. Args: tools: Tool definitions to pass to the LLM, or None to disable tools Re

Periodically check server health and log transitions.

Create and initialize a ToolManager instance. Uses _create_tool_manager() which can be patched via set_tool_manager_factory() for testin

Load scope data from disk. Returns empty file if not found.

Create a MockContext with pending context notices.

Create a minimal ServerInfo for testing.

Extract and register numeric literals from recent user messages. Scans conversation history for the most recent user message(s) and r

Helper to actually run a coroutine passed to asyncio.run.

Safely cleanup UI manager with enhanced error handling. Args: ui: UI manager to cleanup

Save execution checkpoint for resume support.

Add assistant message to conversation history.

Add a message to conversation history.

Add a runtime server to preferences. Args: name: Server name config: Server configuration (command, args, env, transp

Create mock analysis when real servers aren't available.

Send queued notifications that accumulated while WS was down.

Execute tool via stream_manager.

Attempt to fix common array schema issues. Returns: Fixed parameters schema

Get the currently active model.

Get or create a client for the specified provider and model. Args: provider: Provider name model: Model name

Get the singleton DiscoveryManager instance.

Get environment variable as float. Args: var: EnvVar enum member default: Default value if not set or invalid Returns:

Get the environment variable name for this provider's API key.

Get a logger with the given name.