> ## Documentation Index
> Fetch the complete documentation index at: https://praison.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Quality Checking

> Automatic quality assessment for task outputs with memory storage

Tasks with `quality_check=True` score outputs and store high-quality results in memory.

```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Agent, Task, AgentTeam, Memory

agent = Agent(
    name="Writer",
    instructions="Write clear, complete articles.",
    memory=Memory(),
)

task = Task(
    description="Write a short article on AI ethics",
    expected_output="Introduction, three sections, and conclusion",
    agent=agent,
    quality_check=True,
)

AgentTeam(agents=[agent], tasks=[task]).start()
```

The user runs a team task; quality scoring decides whether outputs land in long-term memory.

```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
graph LR
    T[Task output] --> A[Assess]
    A --> S[Score]
    S -->|"> 0.7"| M[Long-term memory]
    S -->|"≤ 0.7"| X[Short-term only]

    classDef task fill:#8B0000,stroke:#7C90A0,color:#fff
    classDef process fill:#189AB4,stroke:#7C90A0,color:#fff
    classDef ok fill:#10B981,stroke:#7C90A0,color:#fff
    classDef skip fill:#F59E0B,stroke:#7C90A0,color:#fff

    class T task
    class A,S process
    class M ok
    class X skip
```

## How It Works

```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
sequenceDiagram
    participant User
    participant Agent
    participant QualityChecking

    User->>Agent: Request
    Agent->>QualityChecking: Process
    QualityChecking-->>Agent: Result
    Agent-->>User: Response
```

## Quick Start

<Steps>
  <Step title="Simple Usage">
    Enable quality checking on a task (default is `True`):

    ```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
    from praisonaiagents import Agent, Task, AgentTeam, Memory

    agent = Agent(name="Writer", instructions="Be thorough.", memory=Memory())

    task = Task(
        description="Explain quantum computing in plain English",
        expected_output="500-word explanation with examples",
        agent=agent,
        quality_check=True,
    )

    AgentTeam(agents=[agent], tasks=[task]).start()
    ```
  </Step>

  <Step title="With Configuration">
    Disable for fast runs or use execution presets:

    ```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
    from praisonaiagents import Agent, Task, AgentTeam

    agent = Agent(name="Draft", instructions="Quick drafts only.")

    task = Task(
        description="Draft a tweet",
        agent=agent,
        quality_check=False,
    )

    AgentTeam(agents=[agent], tasks=[task]).start()
    ```
  </Step>
</Steps>

***

## How It Works

```mermaid theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
sequenceDiagram
    participant User
    participant Agent
    participant Feature as Quality Checking

    User->>Agent: Request
    Agent->>Feature: Process request
    Feature-->>Agent: Result
    Agent-->>User: Response
```

When `quality_check=True` and memory is configured:

1. Agent completes the task
2. `Memory.calculate_quality_metrics()` scores completeness, relevance, clarity, accuracy via LLM
3. `finalize_task_output()` stores in long-term memory only when score exceeds **0.7**
4. Quality metadata attaches to the task result

Memory is required — without it, quality checking logs a warning and skips storage.

***

## Endpoint and Key Resolution

`Memory.calculate_quality_metrics()` picks its scoring endpoint and key from the memory config, so quality scoring reaches the same host you point memory at.

| Path | Endpoint | Key |
| - | - | - |
| litellm (default when installed) | `api_base` ← config `base_url` (or `OPENAI_BASE_URL` / `OPENAI_API_BASE`) | `api_key` ← config `api_key` (or `OPENAI_API_KEY`) |
| openai (fallback) | `base_url` ← config `base_url` (or `OPENAI_BASE_URL`) | `api_key` ← config `api_key` (or `OPENAI_API_KEY`) |

Config keys are read from the top level or a nested `"config"` dict, e.g. `Memory(config={"base_url": ..., "api_key": ...})`.

<Note>
  The litellm path now honours config `base_url` / `api_key` (passed as litellm's `api_base` kwarg). Before [PR #4958](https://github.com/MervinPraison/PraisonAI/pull/4958) it silently went to OpenAI's default endpoint — users pointing memory at Groq, OpenRouter, Together, or a private OpenAI-compatible endpoint will now see the quality-scoring call go to the configured host.
</Note>

```python theme={"theme":{"light":"vitesse-light","dark":"vitesse-dark"}}
from praisonaiagents import Memory

memory = Memory(config={
    "base_url": "https://api.groq.com/openai/v1",
    "api_key": "gsk_...",
})
```

***

## Configuration Options

| Option | Type | Default | Description |
| - | - | - | - |
| `quality_check` | `bool` | `True` | Enable LLM quality assessment |
| `expected_output` | `str` | `None` | Benchmark for scoring (strongly recommended) |
| `memory` | `Memory` | `None` | Required for quality storage |

Execution presets: `"fast"` disables quality check; `"balanced"` and `"thorough"` enable it.

***

## Best Practices

<AccordionGroup>
  <Accordion title="Always set expected_output">
    Clear expectations produce meaningful scores — vague tasks score inconsistently.
  </Accordion>

  <Accordion title="Configure memory first">
    Quality checking stores to memory — attach `memory=Memory()` to the agent or task.
  </Accordion>

  <Accordion title="Disable for drafts">
    Set `quality_check=False` on brainstorming or speed-critical tasks.
  </Accordion>

  <Accordion title="Retrieve high-quality history">
    Search with `min_quality=0.7` to reuse past strong outputs as context.
  </Accordion>
</AccordionGroup>

***

## Related

<CardGroup cols={2}>
  <Card title="Quality-Based RAG" icon="star-half-stroke" href="/docs/features/quality-based-rag">
    Quality scoring for retrieval
  </Card>

  <Card title="Memory" icon="brain" href="/docs/features/memory">
    Memory configuration and search
  </Card>
</CardGroup>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.