# Custom Retriever

<figure><img src="/files/NaBhJ6KK3Wv1zKZn5HJl" alt="" width="298"><figcaption></figcaption></figure>

By default, when context is being retrieved from vector store, they are in the following format:

```json
[ 
    {
        "pageContent": "This is an example",
        "metadata": {
            "source": "example.pdf"
        }
    },
    {
        "pageContent": "This is example 2",
        "metadata": {
            "source": "example2.txt"
        }
    }
]
```

**pageContent** of the array will be joined together as a string, and fed back to LLM for completion.

However, in some cases, you might want to include information from metadata to give more information to LLM, such as source, link, etc. This is where **Custom Retriever** comes in. We can specify the format to return to LLM.

For instance, using the following format:

```javascript
{{context}}
Source: {{metadata.source}}
```

Will results in the combined string as below:

```
This is an example
Source: example.pdf

This is example 2
Source: example2.txt
```

This will be sent back to LLM. Since LLM now has the sources of the answers, we can use prompts to instruct LLM to return answers followed by citations.


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.flowiseai.com/integrations/langchain/retrievers/custom-retriever.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.