Docs · LLMs and developers
Reviewed2026-07-07Version3.12.1Habit performance
How Achiral scores shared habit outcomes, gates low-utility patterns, and lets owners review habit behavior.
Habit performance is the feedback layer for procedural memory.
When a habit template runs, Chiro records the outcome and updates the habit's utility score. That score influences how confidently the habit should run the next time the same pattern appears.
Two learning layers
| Layer | Scope | Purpose |
|---|---|---|
| Organization habit memory | Shared across the organization | Sets the default utility and behavior for a habit |
| Personal habit memory | Private to one member | Adapts the habit to how it works for that person |
Organization outcomes never overwrite a member's private score. A member's personal pause or approval preference never changes the shared organization habit.
How scoring changes behavior
- Ranking: stronger habits can be preferred when several habits match the same trigger.
- Approval gating: weaker habits can require approval before actions execute.
- Suppression: consistently poor habits can be paused from automatic firing during a cooldown.
- Exploration: new or under-tried habits still get a chance before being judged.
This is a reweighting system, not a habit generator. Achiral does not rewrite a habit or invent new ones from the score.
Review surface
Owners and admins review shared habit performance from the organization workspace. The surface shows utility score, success and failure counts, last outcome, and current state.
| State | Meaning |
|---|---|
| Healthy | Scoring well; runs normally |
| Learning | Too few recent runs to judge |
| Approval-gated | Low utility; actions require approval |
| Suppressed | Paused from automatic firing during cooldown |
For private member-level controls, see Your habits.