Skip to content

Playbook · Knowledge · Example output

Example output: a real article becomes a vault note

This is what the knowledge playbook produces. One source worth keeping goes in; a short, sourced note comes out, shown to you before anything is saved.

The inputs

Source: a real article from the Anthropic Privacy Center, "Is my data used for model training?" (live link, checked August 10, 2026). Why keep it: "I use Claude for client work and need to remember what the privacy toggle actually does." Vault layout: notes live in Reference/, lowercase tags. Everything below the line is the note exactly as the playbook delivers it for approval.


The note, as delivered for approval

# Claude training toggle - what it actually controls

tags: #ai #privacy #reference

## What the source says

On personal Claude plans (Free, Pro, Max), Anthropic uses your chats
to improve its models in three cases: you allow it in privacy
settings, a conversation is flagged for safety review, or you
explicitly join a testing program.

Two details worth remembering:

- Incognito chats are not used to improve Claude even when the
  setting is on.
- Turning the setting off stops NEW chats from being used. Data
  already in training that has started stays there.

The setting lives at Settings -> Privacy -> "Help Improve Claude."

## Why it matters to me

Client work goes through Claude. The safe habit: keep the toggle
where I decided, and use Incognito for anything sensitive rather
than trusting memory about the setting.

## Source

Anthropic Privacy Center, "Is my data used for model training?"
https://privacy.claude.com/en/articles/10023580-is-my-data-used-for-model-training
Article dated March 16, 2026. Saved August 10, 2026.

The playbook then asks: "Save to Reference/claude-training-toggle.md?" Nothing is written until you say yes.


Why this output is trustworthy

The summary contains only what the source article says; the "why it matters" section is explicitly the reader's own reasoning, kept separate. The source line carries both the article's date and the save date, so future-you knows exactly how stale it might be. The Before you save it checklist runs at the end.

Want notes like this?

Start the knowledge playbook