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

Kill the cached persistent sandbox if one exists.

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

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

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

Return the active execution uuid, if any.

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

Load a skill at METADATA disclosure level. Parses SKILL.md frontmatter only and validates directory name. Args: skill_dir: Path to t

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

Context manager to temporarily set the platform integration token. Args: integration_token: The integration token to set within the context

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.

Estimate token count using a conservative cross-provider heuristic. Args: text: The text to estimate tokens for. Returns: Es

Invoke after_llm_call hooks for direct LLM calls (no agent context). This method should be called by native provider implementations after

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

Finalize batch and return it for sending

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

Parse a TOML string and return its parsed contents.

Parse tool call arguments from a JSON string or dict. Returns: ``(args_dict, None)`` on success, or ``(None, error_result)`` on J

Build an in-memory .tar.gz containing the given filename → content mapping.

Replace crewai dependency version pins in a pyproject.toml string. Handles both pinned (==) and minimum (>=) version specifiers, as well as e

Fail unless every effective canary CrewAI requirement has the exact pin.

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

Create a deployable ZIP archive for a CrewAI project.

Download a skill from the registry and store it in the cache. Args: org: Organisation slug. name: Skill name. source: Opt

Create a Fingerprint from a dictionary representation. Args: data: Dictionary representation of a fingerprint Returns:

Return the cached skill directory path if usable, else None. Args: org: Organisation slug. name: Skill name.

Get current call_id from context

Handle unknown errors by informing the user. Args: printer: Printer instance for output exception: The exception that occurred

List all CLI configuration parameters.

Context manager that establishes an LLM call scope with a unique call_id.

Load and process all files from a directory recursively. Args: source_content: Directory path or URL to a directory listing

Split SKILL.md content into frontmatter dict and body text. Args: content: Raw SKILL.md file content. Returns: Tuple of (fro

Fetch content from URL (for providers that don't support URL references).

Register a global before_llm_call hook. Global hooks are added to all executors automatically. This is a convenience function for registering

Resolve a registry reference to a Skill object. Resolution order: 1. ./skills/{name}/ in the current working directory (project-local) 2.

Resume Live session updates after human input is complete. New streaming sessions will be created on-demand when needed. This method

Convert to JSON-RPC error object format.

Generate structured output using the specified Pydantic model. Returns: Instance of the specified Pydantic model with structured

Return the bound ``Memory`` or raise a clear error if missing.

Resolve tool specs into tool instances or serialized BaseTool dicts. Strings keep the existing shorthand behavior. Dicts are passed through so

Read a file, or a window of its lines, as text.

Process input text synchronously.

Whether this kickoff should defer final flow trace finalization.

Split messages into chunks at message boundaries. Excludes system messages and expands oversized single messages before chunking. Each chunk

A CrewAI AMP client stub serving one skill.

Add a new task output to storage. Args: task: The task that was executed. output: The output produced by the task.

Create a minimal valid PNG for testing.

Delete an uploaded file. Args: file_id: The file identifier to delete. Returns: True if deletion was success

Execute a single native tool call with full lifecycle management. Handles: arg parsing, tool lookup, max-usage check, cache read/write, befor

Execute *sql_query* and return the results as a list of dicts. Parameters ---------- sql_query: The SQL statement

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

Parse a registry reference into its organisation, name and version. Accepts '@org/skill-name' and the version-pinned '@org/skill-name@1.2.0'

Replay the crew execution from a specific task.

Import a module, optionally returning a specific attribute. Args: name: The module name to import. purpose: Description of what r

Execute the crew and capture the result.

Validate a video file against constraints. Args: file: The video file to validate. constraints: Video constraints to validate aga

Create a simple analyst crew for file analysis.

Create a batch slice from prepared documents. Args: prepared: PreparedDocuments containing ids, texts, and metadatas. start_index

Emit stream chunk event. Args: chunk: The text content of the chunk. from_task: The task that initiated the call.

LiteLLM responses/chunks share the choices-shape with OpenAI/Azure; delegate to the shared extractor.

Find the CheckpointConfig for an event source. Walks known relationships: Task -> Agent -> Crew. Flow and Agent carry their own checkpoint fi

Format messages with role labels for summarization. Skips system messages. Handles None content, tool_calls, and multimodal content blocks.

Apply the executor's stop words to the LLM for the duration of one call. Uses :func:`crewai.llms.base_llm.call_stop_override` so the LLM's stop