Skip to lesson
Exit
Manage chat context and privacy1 / 2

1 min lesson

Reset one chat and verify the controls

Reset one stale chat. Test the ignore rule and check the privacy settings against current docs.

Step 1 of 2

Take one chat that changed goals or repeated a failed fix. Decide whether the goal changed or only the context grew. If it changed, summarize the settled decisions, open a new chat, state the new finish line and attach current evidence. Test a .cursorignoreA file listing paths Cursor must never index or read, kept separate from .gitignore. Press Enter for the full definition. path through Agent and @, then record the terminal and MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. exception. Use .cursorindexingignore only for a file that should stay out of search but remain readable. Finally, verify Privacy ModeCursor's setting that guarantees code data is not used for training by Cursor or its model providers, and that an admin can enforce org-wide; data-retention terms are a separate, contractual layer. Press Enter for the full definition. enforcement, the selected model and the product path before making a retention claim. Record the exact ignore pattern, Privacy Mode setting, selected model and product surface with the handoff so another reviewer can reproduce each claim. Keep the evidence beside the claim.

Learn more

Full explanation

Context ring, ignore files and Privacy Mode

Start a clean chat and check each control
Context ring, ignore files and Privacy Mode
SayThis thread began with an auth refactor, repeated the same broken-import fix twice, and now the task has changed to rate limiting.
SeeThe context ring is nearly full, and the last replies still assume the old task.
TypeSummarize only the auth decisions that the rate-limit task still depends on.
SeeThe summary keeps the token check and middleware order, and drops the failed import attempts.
Type.env.local fixtures/customer-dump.json
DoEdit .cursorignoreA file listing paths Cursor must never index or read, kept separate from .gitignore. Press Enter for the full definition. and add the sensitive paths.
SeeCursor blocks the ignored files from Agent, Tab, Inline Edit and @ mentions. Terminal and MCPModel Context Protocol. A standard that lets an AI agent pull in context from outside the repo, like Jira tickets or internal docs. Press Enter for the full definition. tools are outside this control, so the ignore file is not complete protection.
Typetmp/generated-trace.json
DoAdd the generated trace to .cursorindexingignore because it should stay out of codebase search but remain available to AI features.
SeeThe trace is excluded from indexing only. Agent can still read it when the task needs it.
DoOpen Settings and confirm Privacy ModeCursor's setting that guarantees code data is not used for training by Cursor or its model providers, and that an admin can enforce org-wide; data-retention terms are a separate, contractual layer. Press Enter for the full definition. is enforced for the team. Check whether the selected model or product path has a retention exception.
SeePrivacy ModeCursor's setting that guarantees code data is not used for training by Cursor or its model providers, and that an admin can enforce org-wide; data-retention terms are a separate, contractual layer. Press Enter for the full definition. blocks training and uses ZDRZero Data Retention. A contractual guarantee that the model provider won't store your code or train on it. Press Enter for the full definition. providers by default. Non-ZDR models, personal API keys, abuse review and cloud product retention need their own check.
DoStart a new chat with the short summary and the rate-limit task.
SeeThe new conversation contains the settled auth decisions and current evidence, without the failed import loop.
Learn more

Optional practice

Test yourself on Reset one chat and verify the controls

QWhat should be true when you finish the chat context and privacy controls practice?