Add assistant message with all tool calls to history.

Build a UniversalPlan from a plan dict. Args: plan_dict: Dict with title, steps, tool_calls, dependencies. graph: Gra

Return a side_effect that closes the coroutine then raises *exc*.

Close and remove all file handlers from the root logger.

Create a ToolManager instance using factory or default constructor. This pattern allows tests to inject mock ToolManagers without dynamic getattr

Internal method to clear display (no lock, called with lock held). Clears the current status display. Should only be called when rend

Execute a command with context. Args: command: The unified command to execute. kwargs: Parameters from Typer

Execute a batch of tool calls and append results to messages.

Extract tool name from tool definition.

Fetch HTML content directly from an HTTP/HTTPS URL.

Find an available port starting from _next_port.

Get models from Ollama CLI.

Get or create PlanningContext.

Get the MemoryScopeStore from chat context.

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

Browser client changed which agent it's viewing.

Browser requested list of saved sessions.

Handle SIGINT (Ctrl+C) gracefully.

Test-specific tool initialization.

Initialize tool discovery and adaptation.

Invalidate tool index and LLM tools cache, forcing rebuild on next access.

Create a command with a 'format' parameter that has choices.

Build a mock chat_context with tool_memory.memory.tool_log.

Build a mock ChatContext with the given conversation_history.

Build a fake ToolLogEntry-like object.

Return a mock ToolManager with async helpers wired up.

Create a minimal ToolInfo for testing.

Notify all registered bridges and clean up per-client state.

Register a command group with Typer.

Clear the registry — for testing only.

Resolve timeout from all sources with priority (type-safe!).

Replace token placeholders with actual tokens. Supports two formats: - Legacy: {{token:provider}} - for OAuth tokens - New: $

Resolve a single value, recursing into dicts and lists.

Shared function to run provider commands.

Save scope data to disk with file locking.

Get the file path for a session.

Store a tool result as a VM page so it survives eviction. Without this, tool results (weather forecasts, geocoding data, etc.) exist

Build a mock LLM response with text only (step done). Uses the chuk_llm native format (top-level response key).

Broadcast a message from a specific agent. If no clients have subscriptions configured, broadcasts to all (backward compat). Otherwis

Broadcast a message not scoped to any agent (e.g. AGENT_LIST).

Delete a provider API key from secure storage. Note: This only removes from storage. Environment variables are unaffected. Args:

Disable a tool.

Re-enable a previously disabled tool.

Estimate token count from text using chars/4 heuristic.

Execute a subcommand or the default action.

Execute the health command.

Find a server by name.

Remove a memory entry by key. Returns True if found and removed.

Format all memories as a markdown section for the system prompt. Returns empty string if no entries exist.

Format a human-readable summary of token usage.

Format a tool for display in UI.

Create a display update from current streaming state.

Get all resolved timeouts as immutable config.

Get available models for a provider. NO HARDCODED MODELS - All come from configuration or discovery. Args: prov

Quick function to get available models after discovery. Args: provider: Provider name (uses DEFAULT_PROVIDER if None) Returns:

Get list of all available providers. Returns: List of provider names (sorted alphabetically)

Get the bridge for a running app by its resource URI. Multiple tools can share the same resourceUri (e.g. show_video and play_video b

Get the current application context. Raises: RuntimeError: If context hasn't been initialized

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

Get last used servers.

Get middleware status for diagnostics. Returns Pydantic model dict with retry, circuit breaker, and rate limiting status.

Get recent tool call history from procedural memory.

Get list of currently running apps.

Get the available tool catalog for LLM plan generation. Args: provider: LLM provider format for tool adaptation. Returns

Get verbose setting.

List all provider tokens stored in secure storage. This function checks all known providers from chuk_llm and shows which ones have toke

List all registered commands. Args: mode: Optional mode filter to only list commands that support this mode. Re

List all saved groups. Returns ------- list[dict] List of ``{group_id, description, created_at, agent_count, path}``

Async load runtime configuration. Args: config_path: Path to config file (default: server_config.json) cli_overrides: CLI argumen

Log debug message.

Called when a browser client disconnects. If no clients remain, cancel all pending approval futures so the tool processor doesn't han

Cancel all pending approval futures. Call before stopping server.

Display user message. Args: message: User message to display

Remove tool state for an agent (call when the agent is stopped).

Save the current configuration to file.

Save configuration to JSON file.

Replace all models and update default. Args: models: New list of model IDs

Set whether to confirm tools at a specific risk level. Args: risk_level: 'safe', 'moderate', or 'high' should_confirm

Set server disabled state in preferences. Args: server_name: Name of the server disabled: Whether server should be di

Set up environment variables for ChukLLM discovery. MUST be called before any chuk_llm imports.