Use codemap find "SymbolName" to locate any symbol before reading — it gives exact file + line range and is faster than grep across the whole repo.

Multi-Tool Orchestration & Selection

Quick start

# agent.py — from __future__ import annotations IS safe here (no @tool)
from __future__ import annotations

from lauren_ai import agent, use_tools
from .tools import search_web, query_database, calculate

@agent(model="claude-opus-4-6", system="Use available tools to answer questions.")
@use_tools(search_web, query_database, calculate)
class ResearchAgent: ...

Parallel tool calls

Enable parallel_tool_calls=True to run all tool calls in a single LLM turn concurrently rather than sequentially:

@agent(
    model="claude-opus-4-6",
    system="Use tools efficiently.",
    parallel_tool_calls=True,
)
@use_tools(search_web, query_database, calculate)
class ParallelAgent: ...

Tool error policy

@agent(model="claude-opus-4-6", tool_error_policy="return_error")
@use_tools(search_web, calculate)
class RobustAgent: ...

Testing sequential tool calls

from lauren_ai._transport import Completion, TokenUsage
from lauren_ai._transport._mock import MockTransport

mock = MockTransport()

# LLM calls search_web first
mock.queue_tool_use("search_web", {"query": "latest AI news"})
# Then calls calculate
mock.queue_tool_use("calculate", {"expression": "42 * 2"})
# Final answer
mock.queue_response(Completion(
    id="c3", model="mock-model",
    content="Based on my research: the answer is 84.",
    tool_calls=[], stop_reason="end_turn",
    usage=TokenUsage(input_tokens=30, output_tokens=20),
))

Inspecting tool calls made

resp = await runner.run(ResearchAgent(), "Find recent AI news and compute 42*2")
assert len(resp.tool_calls_made) == 2
assert resp.tool_calls_made[0].name == "search_web"
assert resp.tool_calls_made[1].name == "calculate"

Reference files