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

# Get Messages

> Retrieve raw conversation messages (not extracted memories)

The `get_messages()` method retrieves the original conversation messages stored via `add()`. Unlike `get()` which returns extracted memory records, this returns the raw user/assistant message pairs.

## Basic Usage

```python theme={null}
from lyzr import Cognis

cog = Cognis()

messages = cog.get_messages(
    owner_id="user_alice",
    session_id="sess_001",
)
for msg in messages.get("messages", []):
    print(f"[{msg['role']}] {msg['content']}")
```

## Parameters

| Parameter | Type | Default | Description |
| - | - | - | - |
| `owner_id` | `str` | None | User/tenant scope |
| `session_id` | `str` | None | Session scope |
| `agent_id` | `str` | None | Agent scope |
| `limit` | `int` | None | Max messages to return |
| `latest` | `bool` | `False` | Return only the most recent messages |
| `cross_session` | `bool` | `False` | Include messages from all sessions |

## Use Cases

**Conversation replay** — Show a user their previous conversation:

```python theme={null}
messages = cog.get_messages(
    owner_id="user_alice",
    session_id="sess_001",
    limit=50,
)
```

**Latest context** — Get the most recent messages for prompt injection:

```python theme={null}
messages = cog.get_messages(
    owner_id="user_alice",
    session_id="sess_001",
    latest=True,
    limit=10,
)
```

**Cross-session history** — Pull messages from all sessions:

```python theme={null}
messages = cog.get_messages(
    owner_id="user_alice",
    cross_session=True,
    limit=100,
)
```

## Async

```python theme={null}
messages = await cog.aget_messages(owner_id="user_alice", session_id="sess_001")
```

> [!NOTE]
>
> `get_messages()` is hosted-only. In the open-source `lyzr-cognis` package, raw messages are stored in SQLite and accessible via `get_context()` as part of the short-term memory.


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