save-learning

To save user instructions across sessions, you can trigger the skill by prompting it to "remember this" or apply a rule "from now on." It refines your instruction into an actionable learning and stores it in CALIBER_LEARNINGS.md after confirmation.

The best way to make an AI remember project conventions is to state your rule and ask it to remember the instruction. The skill categorizes the rule as a convention, pattern, or preference, then persists it to CALIBER_LEARNINGS.md for future interactions.

Yes, you can scope persistent instructions to personal preferences by applying the --personal flag. This ensures your saved learnings are categorized as personal context rather than project-wide standards.

Saving persistent learnings works by writing your refined instructions to CALIBER_LEARNINGS.md. After you confirm the categorized learning bullet, the skill stores it in this file to drive consistent behavior in future sessions.

You can persist project standards by asking the skill to remember instructions, conventions, or rules. It refines these into clear bullets categorized as convention, pattern, anti-pattern, preference, or context before storing them.

Yes, user confirmation is required before the skill saves a persistent learning. Once you confirm the refined instruction, it is persisted to CALIBER_LEARNINGS.md to ensure the behavior is applied in future sessions.