MCPcopy Create free account
hub / github.com/ScaleML/AgentSPEX / LoopConfig

Class LoopConfig

src/harness/agentic_loop/session.py:8–20  ·  view source on GitHub ↗

Configuration for agentic loop sessions.

Source from the content-addressed store, hash-verified

6
7@dataclass
8class LoopConfig:
9 """Configuration for agentic loop sessions."""
10
11 model: str = "claude-opus-4-5"
12 mcp_url: str = "http://localhost:7002/mcp"
13 max_turns: int = 200
14 max_tool_calls_per_turn: int = 10
15 temperature: float = 0.2
16 output_dir: str = "outputs/agentic_loop"
17 max_tool_result_chars: int = 30000
18 model_kwargs: dict = field(default_factory=dict)
19 plan_generator_model: str = "gpt-5"
20 plan_executor_model: str = "claude-sonnet-4-6"
21
22
23class Session:

Callers 2

run_loopFunction · 0.85
mainFunction · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected