proof-orchestrator

To standardize a proof idea, this skill creates a clean proof run directory with the correct task and materials files, then performs a preflight difficulty probe to determine if the task is local, needs a rewrite, or should be sent to GPT-pro.

A GPT-pro proof workflow coordinates proof production from task standardization to final assembly. You need audit alignment when packaging handoffs and performing Codex or DeepSeek audits to ensure provenance and review gates are preserved.

Yes, proof-orchestrator applies to proof runs, continuations, stress tests, and bounded repair scenarios. It manages final proof assembly while preserving provenance, review gates, and status labels across these contexts.

Yes, proof-orchestrator packages reviewed prompts, manages GPT-pro output, and performs Codex or DeepSeek audits. It supports bounded local repairs during the handoff and audit flow to maintain proof provenance.

Preparing a reviewed GPT-pro handoff requires reviewed prompt bundles and run-directory state reconstruction. The skill standardizes the request, runs a preflight probe, and packages the handoff without losing proof provenance.

A preflight difficulty probe prevents wasted budget by running a short local check to decide whether a task can be handled locally, needs a rewrite, or should be dispatched to GPT-pro for full proof production.