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

# Personalization

> Teach Foundry how you like to be helped — tone, formatting, default context.

Foundry can be tuned to your style. Today that tuning happens through your **System Prompt**: a static, always-applied instruction that shapes every reply in every chat.

<Note>
  Dynamic **Memory** (Foundry remembering specifics from prior conversations) is on the roadmap — see [Memory](/foundry/personalization/memory).
</Note>

## What the System Prompt does

It's the first thing the model reads, before your message. Use it for:

* **Tone.** *"Reply concisely. Skip pleasantries. Use plain language."*
* **Formatting defaults.** *"Use markdown tables for comparisons. Code blocks for code. No emoji."*
* **Persona.** *"You are a senior staff engineer reviewing a junior's code."*
* **Standing context.** *"I'm building a fintech B2B SaaS. Assume that's the context unless I say otherwise."*

## Where to edit it

**Settings → System Prompt** opens the editor. Type, save, done.

<Note>
  **Screenshot coming.** System Prompt editor.
</Note>

## How it interacts with Skills

Active [Skills](/foundry/skills/overview) layer on top of your System Prompt. Skills are task-specific; the System Prompt is identity-level. Both are sent to the model on every turn.

## Tips

* **Keep it short.** Long system prompts eat token budget every turn. 5–10 lines is plenty for most people.
* **Bullet, don't prose.** Bullets keep the model focused on the rules.
* **Iterate.** When the assistant does something annoying twice, add a one-line rule to fix it.

## Related

<CardGroup cols={2}>
  <Card title="Memory" icon="memory" href="/foundry/personalization/memory">
    Dynamic personalization. Coming soon.
  </Card>

  <Card title="Skills" icon="puzzle-piece" href="/foundry/skills/overview">
    Task-specific behavior on top of System Prompt.
  </Card>
</CardGroup>