Reset (delete all) memories in scope. Args: scope: Scope to reset. If None and root_scope is set, resets only wit

Check if the model supports multimodal inputs. For litellm, check common vision-enabled model prefixes. Returns: True if

Emit LLM call started event.

Run validation with the subprocess-based import check stubbed out; the classifier is exercised directly in its own tests below.

Get all registered global before_tool_call hooks. Returns: List of registered before hooks

Query across all knowledge sources to find the most relevant information. Returns the top_k most relevant chunks. Raises:

Retrieve a prompt by kind and key. Args: kind: The kind of prompt. key: The key of the specific prompt to retrieve.

Summarize messages to fit within context window. Uses structured context compaction: preserves system messages, splits at message boundaries,

Create a simple analyst agent for file analysis.

Create a deep copy of the Agent.

Smart task factory that automatically assigns a mock agent when guardrails are present. This maintains backward compatibility while handling

Retrieve an error message by key. Args: error: The key of the error message to retrieve. Returns: The error

Records that a feature was used. One span = one count. Args: feature: Feature identifier, e.g. "planning:creation",

Return the installed crewAI version string. Falls back to ``"unknown"`` when neither crewai nor crewai-core are pip-installed (e.g. running d

Check if the exception is due to context length exceeding. Args: exception: The exception to check Returns: bool: True if th

Join a root scope with an inner scope, handling edge cases properly. Args: root: The root scope prefix (e.g. '/crew/research-crew').

Context manager for acquiring a read lock. Yields: None

Generic handler for context end events. Args: event_type: Type of the event. source: Source object that triggered the

Sanitize a collection name to meet ChromaDB requirements. Requirements: 1. 3-63 characters long 2. Starts and ends with alphanumeric char

Format text and files as provider-specific multimodal content blocks. This is the main high-level API for converting files to content blocks

Create a Flow instance from a pending feedback state. This classmethod is used to restore a flow that was paused waiting for async hu

Look up a node by event ID. Args: event_id: The event's unique identifier. Returns: The node, or None if not

Get the platform integration token from the current context or environment. Returns: The integration token if set, otherwise None.

Resolve a file to its delivery format for a provider. Args: file: The file to resolve. provider: Provider name (e.g.,

Run a shell command and return output. Args: cmd: Command to run as list of strings. cwd: Working directory for command. Ret

Append a final user-visible assistant message.

Get summary of token usage for this LLM instance. Returns: Dictionary with token usage totals

Store multiple items in memory (non-blocking). The encoding pipeline runs in a background thread. This method returns immediately so

Converts a Python object into a JSON-compatible representation. Supports primitives, datetime objects, collections, dictionaries, dataclasses

Validate that a file path is safe to read. Resolves symlinks and ``..`` components, then checks that the resolved path falls within *base_dir

Context manager for acquiring a write lock. Yields: None

Handle tool execution with proper event emission. Args: function_name: Name of the function to execute function_args:

Prepare parameters for the completion call. Args: messages: Input messages for the LLM tools: Optional list of tool s

Prepare parameters for OpenAI Responses API. The Responses API uses a different structure than Chat Completions: - `input` instead of

Execute the task asynchronously using native async/await.

Create standardized status content with consistent formatting.

Get a todo by its step number.

Register a global after_tool_call hook. Global hooks are added to all tool executions automatically. This is a convenience function for regis

Register a global before_tool_call hook. Global hooks are added to all tool executions automatically. This is a convenience function for regi

Resume flow execution, optionally with human feedback. This method continues flow execution after a flow was paused for async human f

Convert text to pydantic. Args: current_attempt: The current attempt number for conversion retries. Returns:

Create a SkillFrontmatter with the given name.

Executes the tool asynchronously. Args: *args: Positional arguments for the tool. **kwargs: Keyword arguments for the

Call OpenAI API (Chat Completions or Responses based on api setting). Args: messages: Input messages for the completion.

Drain pending saves, flush storage, and shut down the background thread pool.

Flush local shard to central and clean up.

Block until all pending background saves have completed. Called automatically by ``recall()`` and should be called by the crew at shu

Generate a unique document id based on the source reference and content. If the source reference is not provided, the content is used as the s

Return neighbor IDs for a given edge type. Args: edge_type: The relationship type to query. Returns: List of

Save state with a pending feedback marker. This method stores both the flow state and the pending feedback context, allowing the flow

Stop the MCP server.

Create an AzureCompletion(api='responses'). Must be called inside a context where OpenAICompletion is already mocked (i.e. via the ``mock_ope

Emit LLM call failed event.

Return the latest non-prerelease, non-yanked version from PyPI releases.

Format messages for Anthropic API. Anthropic has specific requirements: - System messages are separate from conversation messages

Create a tool from a function. Args: func: The function to create a tool from name: The name of the tool. Defaults to

Load an existing ``Agent`` from a ``.json`` / ``.jsonc`` definition file.

Check if the model supports function calling.

Verify a JWT's signature and claims using PyJWT. Args: jwt_token: The JWT (JWS) string to validate. jwks_url: The URL of the JWKS

Add documents with their embeddings to a collection. Keyword Args: collection_name: The name of the collection to add documents t

Format a message for the LLM. Args: prompt: The message content. role: The role of the message sender, either 'user' or 'assist

Get all registered global after_tool_call hooks. Returns: List of registered after hooks

Generate conversion instructions based on the model and LLM capabilities. Args: model: A Pydantic model class. llm: The language

Load all task output records from the database. Returns: List of dictionaries containing task output records, ordered by task_ind

Pause Live session updates to allow for human input without interference. This stops any active streaming Live session to prevent console ref

Sanitize a name for use in hierarchical scope paths. Converts to lowercase, replaces non-alphanumeric chars (except underscore and hyphen) wi

Sync wrapper for aset.

Wait for all pending event handlers to complete. This helper ensures all sync and async handlers finish processing before proceeding. Useful

Send batch initialization to backend

Return ``(yanked, reason)`` for ``version_str`` against PyPI releases.

Convert a provider usage object to a plain dict and flatten the cache/reasoning sub-counts that LiteLLM nests under provider-specific

Handle and display error messages from API responses. Args: response: The response from the Plus API.

Delete memories matching the given criteria. Args: scope_prefix: Optional scope path prefix. categories: Optional lis

Scan a directory for skill directories containing SKILL.md. Loads each discovered skill at METADATA disclosure level. Args: search_p

Format log record as JSON string.

Get constraints for a provider by name or return if already ProviderConstraints. Args: provider: Provider name string or ProviderConstrai