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

# Memory

> Store and recall long-term memories across runs using Weaviate or pgvector backends

Two-turn demos store then recall user preferences using `WithMemory`. Pick **one backend** per run.

| Backend               | Entrypoint                            |
| --------------------- | ------------------------------------- |
| Weaviate              | `go run ./agent_with_memory/weaviate` |
| PostgreSQL + pgvector | `go run ./agent_with_memory/pgvector` |

Uses the same Docker stack as retriever examples. `task infra:weaviate:up` / `task infra:pgvector:up` creates the **memory** class/table (`AgentMemory` / `agent_memories`) in addition to retriever schema. No seed rows — memories are written by agent runs.

Source: [`examples/agent_with_memory/`](https://github.com/agenticenv/agent-sdk-go/tree/main/examples/agent_with_memory)

## What it demonstrates

* `memory.DefaultConfig` with vector backends
* `MEMORY_STORE_MODE` — `ondemand` (`save_memory` tool) vs `always` (run-end extract)
* Scope via `MEMORY_USER_ID` (default `demo-user`) — must match across runs you want to share
* Telemetry: `total_memory_recalls` / `total_memory_stores` with `SHOW_TELEMETRY=true`

| Backend  | Default store mode |
| -------- | ------------------ |
| Weaviate | `always`           |
| pgvector | `ondemand`         |

Set `MEMORY_RECALL_ENABLED=false` for store-only (skip load before LLM).

## Setup

From `examples/`:

```bash theme={null}
task infra:status    # see what's running
```

**Weaviate:**

```bash theme={null}
task infra:weaviate:up
# EMBEDDING_OPENAI_APIKEY must be in .env before up
go run ./agent_with_memory/weaviate
task infra:weaviate:down   # when finished
```

Verify schema: `curl -s http://localhost:8080/v1/schema | jq '.classes[].class'` — expect `Document` and `AgentMemory`.

**pgvector:**

```bash theme={null}
task infra:pgvector:up
go run ./agent_with_memory/pgvector
task infra:pgvector:down   # when finished
```

After an embedding key change: `task infra:weaviate:down && task infra:weaviate:up`.

## Run behavior

* **No CLI args** — two runs in one process: run 1 stores a preference, run 2 recalls it.
* **With args** — single custom prompt.

```bash theme={null}
go run ./agent_with_memory/weaviate
go run ./agent_with_memory/pgvector "Remember I prefer window seats."
```

Batch all four combinations (each backend × each store mode): `task examples:local` includes memory targets when infra is up.

## Expected output

Two-run flow (no CLI args):

```
[Run 1] Prompt: "Remember I prefer window seats on flights."
[Run 1] I'll remember that you prefer window seats on flights.

[Run 2] Prompt: "What seat do I prefer?"
[Run 2] Based on what you told me earlier, you prefer window seats on flights.
```

Run 1 stores the preference; Run 2 recalls it. With `SHOW_TELEMETRY=true`:

```
--- Telemetry ---
total_memory_recalls: 1
total_memory_stores:  1
```

## Learn more

<CardGroup cols={2}>
  <Card title="Memory" icon="brain" href="/features/memory">
    Store modes, scope, recall config
  </Card>

  <Card title="Retrieval" icon="book-open" href="/examples/retrieval">
    RAG with same vector infra
  </Card>
</CardGroup>