claudex-plan

A planning gate stops execution until a plan survives adversarial review and receives explicit approval. This stop-hook mechanism prevents unsafe rushes to coding during high-stakes tasks like data migrations or multi-service launches.

Prevent rush-to-code by applying a stop-hook gating mechanism to high-stakes tasks like data migrations and security refactors. It requires plans to survive multiple adversarial review rounds before any code is written.

Yes, this stop-hook gating mechanism supports configurable review rounds. The plan must survive adversarial review and receive explicit approval commands across these configured rounds before code may be written.

The planning gate supports both plan-first and plan-during modes. This ensures formal plan survival and explicit approval are enforced whether planning occurs entirely beforehand or concurrently within risky workflows.

Use an adversarial review gate for code approval during high-stakes tasks like multi-service launches, data migrations, and security refactors. It ensures formal plan survival and explicit approval before any code is written.