Review new agents weekly, established ones monthly, and all agents after major context changes
Set agent review cadences at 7 days for new habits, 30 days for established behaviors, and immediately after any major context change, because review timing must match the actual rate of drift in each agent type.
Why This Is a Rule
Behavioral agents drift at different rates depending on maturity. A new agent — installed in the last two weeks — is fragile: it competes against deeply reinforced defaults and can fail silently if the trigger stops being salient, the context shifts, or motivation wanes. Weekly review catches these failure modes before the agent is abandoned. An established agent — firing consistently for 2+ months — is more robust but still subject to context drift: a job change, routine shift, or life event can quietly deactivate it. Monthly review is sufficient to catch gradual drift without over-monitoring.
Major context changes (new job, new home, relationship change, health event) are the most dangerous for all agents because they alter the trigger-condition-action landscape simultaneously. An agent designed for office life may not survive a transition to remote work — not because it's poorly designed but because its triggers don't occur in the new context. Immediate review after any major context change catches these across-the-board disruptions.
The review cadence principle mirrors monitoring in other domains: new deployments get watched closely, stable systems get periodic checks, and any major change triggers immediate inspection.
When This Fires
- When setting up a review schedule for your behavioral agents
- When a new habit feels like it's "sticking" and you're tempted to stop monitoring — keep the weekly review for at least 8 weeks
- After any major life or context change, regardless of how stable your agents seemed before
- When an established agent silently stops firing — you probably need more frequent reviews
Common Failure Mode
Reviewing all agents on the same schedule regardless of maturity. This either over-monitors established agents (creating review fatigue) or under-monitors new ones (missing the critical early failure window). New agents need tight feedback loops; established agents need periodic maintenance checks. Treating them the same wastes attention on the wrong targets.
The Protocol
(1) Categorize each behavioral agent: New (< 8 weeks of consistent firing) or Established (8+ weeks at 80%+ displacement rate). (2) New agents: weekly review. Check displacement rate (Measure behavioral agent progress by displacement rate, not perfection — replacement is gradual, not binary), diagnose failures (Diagnose failing behavioral agents by component — trigger salience, condition scope, or action effort each require different fixes), adjust components as needed. (3) Established agents: monthly review. Verify the agent is still firing, the context hasn't shifted, and the behavior is still producing desired outcomes. (4) After any major context change: review ALL agents immediately, regardless of maturity. Ask: "Does this agent's trigger still occur in my new context? Does the condition still make sense? Is the action still appropriate?" (5) Rebuild agents that don't survive the context change rather than trying to force-fit them to the new environment.