Run the mandatory Qubex verification stack when changes affect runtime code, tests, examples, packaging, or build/test behavior. Use after substantive code changes, and do not mark the work complete until the required checks pass in the `uv` environment.
Audit Qubex documentation against code changes and report missing, incorrect, or outdated docs. Use when public APIs, behavior, setup, dependencies, examples, release notes, or developer guidance change; by default, inspect and prioritize the required updates first, then edit only after the user asks for docs changes.
Run Qubex examples in non-interactive auto mode, capture logs, and provide exact rerun commands. Use when validating notebooks or tutorial examples after code, docs, or dependency changes, especially for simulator and other offline examples; use the bundled script for execution, logging, and rerun-file generation instead of rebuilding the shell recipe by hand.
Compare the current Qubex release candidate against the previous tag and review release readiness. Use before beta, rc, or GA cuts to check version drift, release notes, compatibility messaging, build health, and validation status; use the bundled script to resolve the comparison range and collect the release diff mechanically.
Choose the Qubex compatibility boundary and implementation approach before editing runtime code, public APIs, imports, config formats, or behavior. Use before making API or runtime changes, especially when you must decide between direct cleanup and backward-compatibility handling.
Pull current OpenAI API and platform documentation through the official docs workflow. Use when a task depends on current OpenAI models, APIs, SDKs, or platform behavior and memory could be stale.
Create a branch suggestion, PR title, and draft description after substantive Qubex changes are finished. Trigger when wrapping up a moderate-or-larger change that touched runtime code, tests, build config, or docs with behavior impact and you need a PR-ready handoff block with summary, compatibility notes, and verification results.
Run Qubex coverage, identify the highest-value gaps, and propose focused tests. Use when improving test depth after code changes, debugging weakly covered behavior, or preparing for release hardening; by default, inspect coverage artifacts and recommend the best test additions before editing tests.