CIRK Team Workflow — Adopting CIRK in Your Team

Team workflow

CIRK works best when every task receives a score before execution starts.

A consistent workflow ensures agents and humans follow the same rules.

Any unit of work enters the backlog or queue.

Score C, I, R, K (1–3 each) before execution begins.

Autonomous, guided, draft-first, supervised, or blocked.

The policy applies regardless of who or what executes the work.

Track results and refine scoring over time. Learn about calibration →

Policy interpretation

Low R

More autonomy is acceptable

High R

Human review is mandatory

High K

Rollout needs stronger control

High I

Draft-first behavior is preferred

What CIRK replaces

× Vague estimation

× Inconsistent review expectations

× Unclear autonomy boundaries

× Hidden deployment risk

Shared language

Once a team uses CIRK consistently, people start speaking in a shared operational language.

"This is high R."

"This is low K, we can ship safely."

"This looks like I3, let the agent draft first."

"This is high C, prepare context before execution."

That shared language is what makes a standard adoptable.