Astra-Ares
A Codex hook that asks Jev to choose GPT-6 reasoning effort and how many generations to keep that choice.
Added to Jevfast
How it uses Jev
Bridge passes bounded task context, validates two typed choices against model capabilities, applies effort, then schedules reassessment.
What Jev decides
Illustrative reasoning effort and lease requestExample answers · not a recorded Jev response · Source ↗
Question 1 · effort
YOUR APP
INSTRUCTION
Select the lowest supported reasoning effort sufficient for the next generation based on task and unresolved work.
STATE
Current Codex task state plus the selected model supported-effort list. Representative model supports low, medium and high, with maxLeaseSteps=10.
JEV · CHOICE
- low
- medium
- high
Question 2 · lease
YOUR APP
INSTRUCTION
Select upcoming generation count for which reasoning requirement is likely stable.
STATE
Current Codex task state plus the selected model supported-effort list. Representative model supports low, medium and high, with maxLeaseSteps=10.
JEV · CHOICE
- 1
- 2
- 5
- 10
App workflow
- Gather task context
The hook collects current task state, model capabilities and recent tool results.
- Choose effort and lease
Jev selects supported reasoning effort and the number of generations before reassessment. The bridge validates both choices before applying them.
- Reassess
Lease expiry, new user input, a failed tool, model change or manual effort change triggers another assessment.
Why it is interesting
Turns reasoning depth into an adaptive control loop with explicit reassessment triggers.