# Update Crews

Endpoint: PUT /crewai-agents/api/v1/crews/{crew_id}
Version: 1.0.0

## Path parameters:

  - `crew_id` (any, required)

## Header parameters:

  - `X-Tenant-Id` (any, required)

  - `X-User-Id` (any, required)

## Request fields (application/json):

  - `name` (any, required)
    Name of the overall crew configuration or scenario.

  - `description` (any, required)
    Summary of the crew’s purpose and functionality.

  - `agents` (array, required)
    List of agent definitions in this crew.

  - `agents.name` (any, required)
    Unique name of the agent.

  - `agents.role` (any, required)
    Title or function of the agent within the Crew.

  - `agents.goal` (any, required)
    Primary objective or responsibility of the agent.

  - `agents.backstory` (any)
    Optional narrative to give context and personality to the agent.

  - `agents.tools` (any)
    List of tools this agent is allowed to use.

  - `agents.llm` (any)
    The language model used to power the agent.

  - `agents.max_iter` (any)
    Maximum number of reasoning iterations allowed during task execution.

  - `agents.reasoning` (any)
    If true, the agent will perform reasoning steps before executing tasks.

  - `agents.max_reasoning_steps` (any)
    Maximum number of reasoning steps the agent can take before executing a task.

  - `agents.allow_code_execution` (any)
    If true, the agent can execute code as part of its reasoning process.

  - `agents.code_execution_mode` (any)
    Determines when the agent can execute code: 'safe' (default), or 'unsafe'.

  - `agents.max_retry_limit` (any)
    Maximum number of retries allowed for task execution in case of failures.

  - `agents.knowledge_bases` (any)
    List of knowledge bases that the agent can access.

  - `tasks` (array, required)
    List of task definitions to be executed by the agents.

  - `tasks.name` (any, required)
    Unique identifier for the task.

  - `tasks.description` (any, required)
    Instructions or objectives for the task.

  - `tasks.agent` (any, required)
    Name of the agent responsible for executing the task.

  - `tasks.expected_output` (any)
    Description of the desired output format or structure.

  - `tasks.async_execution` (any)
    If true, the task can be executed asynchronously.

  - `tasks.tools` (any)
    List of tools specific to this task.

  - `tasks.context` (array)
    List of prior task names whose outputs are needed as context for this task.

  - `tasks.human_input` (any)
    If true, the task expects manual human input or approval.

  - `tasks.markdown` (any)
    If true, the task output should be formatted in Markdown.

  - `tasks.output_format` (any)
    Format of the task output: 'markdown' or 'json'.

  - `tasks.output_json_structure` (any)
    Dictionary defining the structure of the expected JSON output, if output_format is 'json'.

  - `tasks.output_json_schema` (any)
    JSON schema defining the structure of the expected output, if output_format is 'json'.

  - `process` (any)
    Execution strategy: 'sequential' or 'hierarchical'.
    Enum: "sequential", "hierarchical"

  - `function_calling_llm` (any)
    LLM used for any function-calling tasks or operations.

  - `manager_agent` (any)
    Name of the agent acting as the manager.

  - `manager_llm` (any)
    LLM used specifically for the manager agent.

  - `planning` (any)
    Enables dynamic planning of task execution by the manager agent.

  - `planning_llm` (any)
    LLM used specifically for dynamic planning.

## Response 422 fields (application/json):

  - `detail` (array)

  - `detail.loc` (array, required)

  - `detail.msg` (any, required)

  - `detail.type` (any, required)