mirror of
https://github.com/crewAIInc/crewAI.git
synced 2026-05-07 18:19:00 +00:00
60 lines
2.0 KiB
Plaintext
60 lines
2.0 KiB
Plaintext
---
|
|
title: Reasoning
|
|
description: "Agent reasoning has been renamed to planning_config. See the Planning page for the current API."
|
|
icon: brain
|
|
mode: "wide"
|
|
---
|
|
|
|
## Overview
|
|
|
|
<Warning>
|
|
The `reasoning=True` and `max_reasoning_attempts=N` arguments on `Agent` are **deprecated**. They still work for now — passing them emits a `DeprecationWarning` and CrewAI auto-migrates the values into a `PlanningConfig` — but they will be removed in a future release.
|
|
|
|
The replacement is **`planning_config`**, documented in full on the [Planning](/en/concepts/planning) page.
|
|
</Warning>
|
|
|
|
## Migration
|
|
|
|
The new API lives on `Agent.planning_config` and uses the `PlanningConfig` model. The presence of a `PlanningConfig` enables planning — there is no separate boolean flag.
|
|
|
|
<CodeGroup>
|
|
```python Before (deprecated)
|
|
from crewai import Agent
|
|
|
|
agent = Agent(
|
|
role="Data Analyst",
|
|
goal="Analyze data and provide insights",
|
|
backstory="Expert data analyst.",
|
|
reasoning=True,
|
|
max_reasoning_attempts=3,
|
|
)
|
|
```
|
|
|
|
```python After
|
|
from crewai import Agent, PlanningConfig
|
|
|
|
agent = Agent(
|
|
role="Data Analyst",
|
|
goal="Analyze data and provide insights",
|
|
backstory="Expert data analyst.",
|
|
planning_config=PlanningConfig(max_attempts=3),
|
|
)
|
|
```
|
|
</CodeGroup>
|
|
|
|
Field mapping:
|
|
|
|
- `reasoning=True` → presence of `planning_config` enables planning.
|
|
- `max_reasoning_attempts=N` → `PlanningConfig(max_attempts=N)`.
|
|
|
|
## What's New
|
|
|
|
`PlanningConfig` exposes capabilities that the old `reasoning` flag did not, including:
|
|
|
|
- `reasoning_effort` (`"low"` / `"medium"` / `"high"`) to control post-step observation, replanning, and refinement.
|
|
- `max_steps`, `max_replans`, `max_step_iterations`, and `step_timeout` to bound plan size and execution.
|
|
- A dedicated planning `llm` separate from the agent's execution LLM.
|
|
- Custom `system_prompt`, `plan_prompt`, and `refine_prompt` overrides.
|
|
|
|
For the full field reference, the Plan-and-Act loop, and guidance on when to use agent-level planning vs. crew-level planning, see [Planning](/en/concepts/planning).
|