跳转至

AgentConfig

ToolkitConfig

Bases: ConfigBaseModel

Toolkit config.

源代码位于： utu/config/agent_config.py 
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
class ToolkitConfig(ConfigBaseModel):
    """Toolkit config."""

    mode: Literal["builtin", "customized", "mcp"] = "builtin"
    """Toolkit mode."""
    env_mode: Literal["local", "e2b"] = "local"
    """Environment mode for the toolkit."""
    name: str | None = None
    """Toolkit name."""
    activated_tools: list[str] | None = None
    """Activated tools, if None, all tools will be activated."""
    config: dict | None = Field(default_factory=dict)
    """Specified  configs for certain toolkit. We use raw dict for simplicity"""
    config_llm: ModelConfigs | None = None  # | dict[str, ModelConfigs]
    """LLM config if used in toolkit."""
    customized_filepath: str | None = None
    """Customized toolkit filepath."""
    customized_classname: str | None = None
    """Customized toolkit classname."""
    mcp_transport: Literal["stdio", "sse", "streamable_http"] = "stdio"
    """MCP transport."""
    mcp_client_session_timeout_seconds: int = 20
    """The read timeout passed to the MCP ClientSession. We set it bigger to avoid timeout expections."""

mode class-attribute instance-attribute 

mode: Literal['builtin', 'customized', 'mcp'] = 'builtin'

Toolkit mode.

env_mode class-attribute instance-attribute 

env_mode: Literal['local', 'e2b'] = 'local'

Environment mode for the toolkit.

name class-attribute instance-attribute 

name: str | None = None

Toolkit name.

activated_tools class-attribute instance-attribute 

activated_tools: list[str] | None = None

Activated tools, if None, all tools will be activated.

config class-attribute instance-attribute 

config: dict | None = Field(default_factory=dict)

Specified configs for certain toolkit. We use raw dict for simplicity

config_llm class-attribute instance-attribute 

config_llm: ModelConfigs | None = None

LLM config if used in toolkit.

customized_filepath class-attribute instance-attribute 

customized_filepath: str | None = None

Customized toolkit filepath.

customized_classname class-attribute instance-attribute 

customized_classname: str | None = None

Customized toolkit classname.

mcp_transport class-attribute instance-attribute 

mcp_transport: Literal['stdio', 'sse', 'streamable_http'] = 'stdio'

MCP transport.

mcp_client_session_timeout_seconds class-attribute instance-attribute 

mcp_client_session_timeout_seconds: int = 20

The read timeout passed to the MCP ClientSession. We set it bigger to avoid timeout expections.

AgentConfig

Bases: ConfigBaseModel

Overall agent config

源代码位于： utu/config/agent_config.py 
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
class AgentConfig(ConfigBaseModel):
    """Overall agent config"""

    type: Literal["simple", "orchestra", "orchestrator", "workforce"] = "simple"
    """Agent type"""

    # simple agent config
    model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Model config, with model_provider, model_settings, model_params"""
    agent: ProfileConfig = Field(default_factory=ProfileConfig)
    """Agent profile config"""
    context_manager: ContextManagerConfig = Field(default_factory=ContextManagerConfig)
    """Context manager config"""
    env: EnvConfig = Field(default_factory=EnvConfig)
    """Env config"""
    toolkits: dict[str, ToolkitConfig] = Field(default_factory=dict)
    """Toolkits config"""
    max_turns: int = 50
    """Max turns for simple agent. This param is derived from @openai-agents"""
    stop_at_tool_names: list[str] | None = None
    """Stop at tools for simple agent. This param is derived from @openai-agents"""

    # orchestra agent config
    planner_model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Planner model config"""
    planner_config: dict = Field(default_factory=dict)
    """Planner config (dict)\n
    - `examples_path`: path to planner examples json file"""
    workers: dict[str, "AgentConfig"] = Field(default_factory=dict)
    """Workers config"""
    workers_info: list[dict] = Field(default_factory=list)
    """Workers info, list of {name, desc, strengths, weaknesses}\n
    - `name`: worker name
    - `desc`: worker description
    - `strengths`: worker strengths
    - `weaknesses`: worker weaknesses"""
    reporter_model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Reporter model config"""
    reporter_config: dict = Field(default_factory=dict)
    """Reporter config (dict)\n
    - `template_path`: template Jinja2 file path, with `question` and `trajectory` variables"""

    # workforce agent config
    workforce_planner_model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Workforce planner model config"""
    workforce_planner_config: dict = Field(default_factory=dict)
    """Workforce planner config (dict)"""
    workforce_assigner_model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Workforce assigner model config"""
    workforce_assigner_config: dict = Field(default_factory=dict)
    """Workforce assigner config (dict)"""
    workforce_answerer_model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Workforce answerer model config"""
    workforce_answerer_config: dict = Field(default_factory=dict)
    """Workforce answerer config (dict)"""
    workforce_executor_agents: dict[str, "AgentConfig"] = Field(default_factory=dict)
    """Workforce executor agents config"""
    workforce_executor_config: dict = Field(default_factory=dict)
    """Workforce executor config (dict)"""
    workforce_executor_infos: list[dict] = Field(default_factory=list)
    """Workforce executor infos, list of {name, desc, strengths, weaknesses}"""

    # orchestrator agent config
    orchestrator_router: "AgentConfig" = None
    """Orchestrator router agent config"""
    orchestrator_config: dict = Field(default_factory=dict)
    """Orchestrator config (dict)\n
    - `name`: name of the orchestrator-workers system
    - `examples_path`: path to planner examples. default utu/data/plan_examples/chain.json
    - `additional_instructions`: additional instructions for planner
    - `add_chitchat_subagent`: whether to add chitchat subagent. default True"""
    orchestrator_model: ModelConfigs = Field(default_factory=ModelConfigs)
    """Planner model config"""
    orchestrator_workers: dict[str, "AgentConfig"] = Field(default_factory=dict)
    """Workers config"""
    orchestrator_workers_info: list[dict] = Field(default_factory=list)
    """Workers info, list of {name, description}"""

type class-attribute instance-attribute 

type: Literal['simple', 'orchestra', 'orchestrator', 'workforce'] = 'simple'

Agent type

model class-attribute instance-attribute 

model: ModelConfigs = Field(default_factory=ModelConfigs)

Model config, with model_provider, model_settings, model_params

agent class-attribute instance-attribute 

agent: ProfileConfig = Field(default_factory=ProfileConfig)

Agent profile config

context_manager class-attribute instance-attribute 

context_manager: ContextManagerConfig = Field(default_factory=ContextManagerConfig)

Context manager config

env class-attribute instance-attribute 

env: EnvConfig = Field(default_factory=EnvConfig)

Env config

toolkits class-attribute instance-attribute 

toolkits: dict[str, ToolkitConfig] = Field(default_factory=dict)

Toolkits config

max_turns class-attribute instance-attribute 

max_turns: int = 50

Max turns for simple agent. This param is derived from @openai-agents

stop_at_tool_names class-attribute instance-attribute 

stop_at_tool_names: list[str] | None = None

Stop at tools for simple agent. This param is derived from @openai-agents

planner_model class-attribute instance-attribute 

planner_model: ModelConfigs = Field(default_factory=ModelConfigs)

Planner model config

planner_config class-attribute instance-attribute 

planner_config: dict = Field(default_factory=dict)

Planner config (dict)

  • examples_path: path to planner examples json file

workers class-attribute instance-attribute 

workers: dict[str, AgentConfig] = Field(default_factory=dict)

Workers config

workers_info class-attribute instance-attribute 

workers_info: list[dict] = Field(default_factory=list)

Workers info, list of {name, desc, strengths, weaknesses}

  • name: worker name
  • desc: worker description
  • strengths: worker strengths
  • weaknesses: worker weaknesses

reporter_model class-attribute instance-attribute 

reporter_model: ModelConfigs = Field(default_factory=ModelConfigs)

Reporter model config

reporter_config class-attribute instance-attribute 

reporter_config: dict = Field(default_factory=dict)

Reporter config (dict)

  • template_path: template Jinja2 file path, with question and trajectory variables

workforce_planner_model class-attribute instance-attribute 

workforce_planner_model: ModelConfigs = Field(default_factory=ModelConfigs)

Workforce planner model config

workforce_planner_config class-attribute instance-attribute 

workforce_planner_config: dict = Field(default_factory=dict)

Workforce planner config (dict)

workforce_assigner_model class-attribute instance-attribute 

workforce_assigner_model: ModelConfigs = Field(default_factory=ModelConfigs)

Workforce assigner model config

workforce_assigner_config class-attribute instance-attribute 

workforce_assigner_config: dict = Field(default_factory=dict)

Workforce assigner config (dict)

workforce_answerer_model class-attribute instance-attribute 

workforce_answerer_model: ModelConfigs = Field(default_factory=ModelConfigs)

Workforce answerer model config

workforce_answerer_config class-attribute instance-attribute 

workforce_answerer_config: dict = Field(default_factory=dict)

Workforce answerer config (dict)

workforce_executor_agents class-attribute instance-attribute 

workforce_executor_agents: dict[str, AgentConfig] = Field(default_factory=dict)

Workforce executor agents config

workforce_executor_config class-attribute instance-attribute 

workforce_executor_config: dict = Field(default_factory=dict)

Workforce executor config (dict)

workforce_executor_infos class-attribute instance-attribute 

workforce_executor_infos: list[dict] = Field(default_factory=list)

Workforce executor infos, list of {name, desc, strengths, weaknesses}

orchestrator_router class-attribute instance-attribute 

orchestrator_router: AgentConfig = None

Orchestrator router agent config

orchestrator_config class-attribute instance-attribute 

orchestrator_config: dict = Field(default_factory=dict)

Orchestrator config (dict)

  • name: name of the orchestrator-workers system
  • examples_path: path to planner examples. default utu/data/plan_examples/chain.json
  • additional_instructions: additional instructions for planner
  • add_chitchat_subagent: whether to add chitchat subagent. default True

orchestrator_model class-attribute instance-attribute 

orchestrator_model: ModelConfigs = Field(default_factory=ModelConfigs)

Planner model config

orchestrator_workers class-attribute instance-attribute 

orchestrator_workers: dict[str, AgentConfig] = Field(default_factory=dict)

Workers config

orchestrator_workers_info class-attribute instance-attribute 

orchestrator_workers_info: list[dict] = Field(default_factory=list)

Workers info, list of {name, description}