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

# Memory

> Foggy proposes facts it learns during investigations. You approve them to keep.

While investigating, Foggy notices facts about your system that would help next time — "the checkout service uses Redis for session cache", "DNS failures here are usually CoreDNS OOMs". These land as pending **Memory** entries for your review.

## How it works

1. During an investigation, Foggy proposes a memory when it learns something worth remembering.
2. The proposal appears in **Knowledge Base → Memory** as a pending item.
3. **Accept**, **Edit**, or **Reject**.
4. Accepted memories become regular Knowledge Base entries and are included in the next investigation — same as a manually written runbook.

Only user-approved memories are used. Foggy cannot write or modify memory on its own.

## Relation to Knowledge Base

Accepted memories live in the same table as Product Docs and Runbooks, so they behave identically — toggleable, editable, included as context. The only difference is origin: memories are proposed by Foggy; Knowledge Base entries are written by you.

## Next steps

<CardGroup cols={2}>
  <Card title="Knowledge Base" icon="book" href="/features/knowledge-base">
    The Product Docs and Runbooks you write manually.
  </Card>

  <Card title="Automations" icon="clock" href="/features/automations">
    Scheduled and alert-triggered investigations.
  </Card>
</CardGroup>