Correction Loop and Reflection Penalty Estimator

Estimate worst-case cost escalation during iterative tool re-tries and autonomous agent correction runs.

Worst-Case Reflection Cost Formula:

$ \text{Cost}_{\text{worst}} = \text{Base Step Cost} \times N_{\text{max iterations}} \times \Gamma_{\text{loop}} $

Reflection Parameters

Base Single-Step Cost ($)

Max Retries / Iteration Limit ($N_{\text{max}}$)

History Context Expansion Factor ($\Gamma_{\text{loop}}$)

Base Single Step $0.0200

Worst Case Reflection Cost $0.1200