# Agno Reasoning — Built-In Step-by-Step Tool Use

> Agno Reasoning gives any Agent built-in chain-of-thought + tool use. Set reasoning=True; the agent plans, calls tools, returns answers with audit.

## Install

Copy the content below into your project:

## Quick Use

1. `pip install agno anthropic duckduckgo-search`
2. Create an Agent with `reasoning=True` plus tools
3. Call `agent.print_response(query, show_full_reasoning=True)` to see each step

---

## Intro

Agno Reasoning is the built-in chain-of-thought + tool-use loop. Set `reasoning=True` on any Agno Agent — internally Agno runs a plan-act-reflect cycle, calls tools, and returns the final answer plus an audit trail of every step. Best for: production agents where you need explainable answers, not just outputs. Works with: Agno 1.0+, any LLM via LiteLLM. Setup time: 1 minute.

---

### Enable reasoning

```python
from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.tools.duckduckgo import DuckDuckGoTools
from agno.tools.calculator import CalculatorTools

agent = Agent(
    model=Claude(id="claude-3-5-sonnet-20241022"),
    tools=[DuckDuckGoTools(), CalculatorTools()],
    reasoning=True,
    reasoning_min_steps=3,    # take at least 3 reasoning steps
    reasoning_max_steps=10,   # cap to prevent runaway
    show_tool_calls=True,
    markdown=True,
)

agent.print_response(
    "Compare the energy density of LFP and NMC batteries in Wh/kg, then "
    "calculate how many extra km a Tesla Model 3 with NMC vs LFP would get.",
    show_full_reasoning=True,
)
```

The output shows each reasoning step:

```
[Step 1: Plan] I need: (1) energy density of LFP, (2) of NMC, (3) Model 3 weight, (4) calculation.
[Step 2: Search] DuckDuckGoTools(query="LFP battery energy density Wh/kg")
[Step 3: Search] DuckDuckGoTools(query="NMC battery energy density Wh/kg")
[Step 4: Calc] CalculatorTools(...)
[Step 5: Reflect] Numbers consistent. Producing final answer.

Final answer:
- LFP: ~150 Wh/kg
- NMC: ~250 Wh/kg
- Model 3 ~80 km extra range with NMC
```

### When to enable reasoning vs leave it off

| Use reasoning | Skip reasoning |
|---|---|
| Multi-step research, math, planning | One-shot Q&A |
| Tool-use over 3+ tools | Fixed prompt template |
| Debugging "why did the agent answer X" | Latency-critical chat |
| Compliance / audit needed | Latency < 1s required |

Reasoning adds ~2-10x to runtime depending on `reasoning_max_steps`, so enable selectively per task class.

---

### FAQ

**Q: Is Agno free?**
A: Yes — Agno is open-source under MPL-2.0. The framework, Agents, Reasoning, Memory, Knowledge are all free. Agno Cloud (managed monitoring) is paid.

**Q: Reasoning vs OpenAI o1 — what's the difference?**
A: o1 is a reasoning model with hidden CoT. Agno Reasoning is a framework feature that wraps any model in an explicit reasoning loop with tool use. You can layer them: use o1 as the model AND enable Agno Reasoning for tool orchestration.

**Q: How does this differ from Agno's plain Agent?**
A: A plain Agent calls tools when the model decides. Reasoning forces a structured plan-act-reflect cycle with min_steps and max_steps you control — better for traceability, slightly higher latency.

---

## Source & Thanks

> Built by [Agno](https://github.com/agno-agi). Licensed under MPL-2.0.
>
> [agno-agi/agno](https://github.com/agno-agi/agno) — ⭐ 22,000+

---

<!-- ZH -->

## 快速使用

1. `pip install agno anthropic duckduckgo-search`
2. 创建 Agent，设 `reasoning=True` 加工具
3. 调 `agent.print_response(query, show_full_reasoning=True)` 看每一步

---

## 简介

Agno Reasoning 是内置的 chain-of-thought + 工具使用循环。任意 Agno Agent 上设 `reasoning=True` —— 内部跑「规划-执行-反思」循环、调工具、返回最终答案 + 完整步骤审计。适合需要可解释答案而不只是输出的生产 agent。需要 Agno 1.0+，任意 LLM（走 LiteLLM）。装机时间 1 分钟。

---

### 开启 reasoning

```python
from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.tools.duckduckgo import DuckDuckGoTools
from agno.tools.calculator import CalculatorTools

agent = Agent(
    model=Claude(id="claude-3-5-sonnet-20241022"),
    tools=[DuckDuckGoTools(), CalculatorTools()],
    reasoning=True,
    reasoning_min_steps=3,    # 至少 3 个推理步
    reasoning_max_steps=10,   # 上限防失控
    show_tool_calls=True,
    markdown=True,
)

agent.print_response(
    "Compare the energy density of LFP and NMC batteries in Wh/kg, then "
    "calculate how many extra km a Tesla Model 3 with NMC vs LFP would get.",
    show_full_reasoning=True,
)
```

输出显示每一步推理：

```
[Step 1: Plan] I need: (1) energy density of LFP, (2) of NMC, (3) Model 3 weight, (4) calculation.
[Step 2: Search] DuckDuckGoTools(query="LFP battery energy density Wh/kg")
[Step 3: Search] DuckDuckGoTools(query="NMC battery energy density Wh/kg")
[Step 4: Calc] CalculatorTools(...)
[Step 5: Reflect] Numbers consistent. Producing final answer.

Final answer:
- LFP: ~150 Wh/kg
- NMC: ~250 Wh/kg
- Model 3 ~80 km extra range with NMC
```

### 什么时候开 reasoning vs 关

| 开 reasoning | 不用 |
|---|---|
| 多步研究、数学、规划 | 一次性问答 |
| 用 3+ 个工具 | 固定 prompt 模板 |
| 排查"agent 为啥这么答" | 延迟敏感聊天 |
| 需要合规 / 审计 | 延迟 < 1s 必需 |

Reasoning 按 `reasoning_max_steps` 增加 ~2-10 倍运行时间，按任务类型选择性开启。

---

### FAQ

**Q: Agno 免费吗？**
A: 免费 —— MPL-2.0 开源。框架、Agent、Reasoning、记忆、知识全免费。Agno Cloud（托管监控）付费。

**Q: Reasoning 跟 OpenAI o1 啥区别？**
A: o1 是推理模型，CoT 是隐藏的。Agno Reasoning 是框架特性，把任何模型包进显式推理循环 + 工具使用。可以叠加：用 o1 当模型 + 开 Agno Reasoning 做工具编排。

**Q: 跟普通 Agent 啥区别？**
A: 普通 Agent 在模型决定时才调工具。Reasoning 强制结构化的「规划-执行-反思」循环，min_steps 和 max_steps 你控制 —— 可追踪性更好，延迟略高。

---

## 来源与感谢

> Built by [Agno](https://github.com/agno-agi). Licensed under MPL-2.0.
>
> [agno-agi/agno](https://github.com/agno-agi/agno) — ⭐ 22,000+


---
Source: https://tokrepo.com/en/workflows/agno-reasoning-built-in-step-by-step-tool-use
Author: Agno