provider-openai-proxy-user-corrections-2

To turn recurring user corrections into reusable rules, you capture repetitive feedback across agent decisions and prompts, extract it as explicit rules, and store the guidance for future reuse to ensure consistency.

Stop repetitive feedback inconsistencies by extracting user corrections as concrete rules, propagating them across ongoing tasks, and revalidating outputs against the stored guidance to reduce rework.

Rule extraction works by capturing recurring user corrections as explicit policies, applying them to current work, revalidating the outputs, and storing the guidance to drive continuous learning across workflows.

Yes, you can apply extracted correction rules automatically to ongoing agent decisions and prompts, ensuring that current work and related outputs are revalidated for consistency without manual intervention.

No dependencies are required to automate extracting corrections as rules, as the pattern operates natively within OpenCode workflows to capture feedback, propagate rules, and store guidance for reuse.

Avoid using continuous-learning rule extraction for one-off, non-recurring feedback, as the overhead of extracting, propagating, and storing guidance only pays off when correcting repetitive issues across multiple tasks.