LearnAIAgents

REMIT Reflection Worksheet

Five questions before every deployment: Responsibility · Envelope · Monitoring · Identity · Trust.

R
Responsibility
Who is accountable when this agent acts?
Name a person, not a team. Define the escalation chain. No orphans.
E
Envelope
What are the hard boundaries of its authority?
Tools · Actions · Spend · Scope. In CODE, not just policy.
M
Monitoring
How will you observe and audit its behaviour?
Action logs, reasoning traces, drift detection, alerts, circuit breakers, approval-fatigue watch.
I
Identity
What is its verified provenance?
Model + version + provider. Capabilities. Who authorised it. If it calls other agents — their identities.
T
Trust
What level of autonomy has been earned?
Intern · Junior · Senior · Principal. Promotion requires evidence. Demotion is always available.