Ask Claude to challenge a change, plan, or prior Claude conclusion in read-only mode, including a resumed follow-up. Use for adversarial review of assumptions, rollback, races, reliability, security, and user impact.
Explore divergent options with Claude in read-only mode. Use for consequential architecture, design, debugging, migration, security, or release choices needing fresh alternatives.
Ask Claude for a read-only second opinion or a follow-up in a compatible prior Claude session, then synthesize it critically. Use for consequential architecture, design, debugging, migration, security, or release decisions.
Delegate diagnosis, implementation, and verification to Claude. Use when the user wants Claude to solve or rescue difficult coding work, in foreground or background.
Retrieve a repository-scoped Claude companion result. Use when a foreground or background job has finished and the user wants output, errors, or changed-file evidence.
Have Claude perform a read-only code review. Use for current uncommitted work or a requested base ref when findings, risks, and precise locations are needed.
Verify Claude Code prerequisites and authentication. Use when installing the plugin, diagnosing setup, or checking whether subscription, API, or enterprise login is ready.
Create a fresh read-only semantic handoff of current Codex context to a new resumable Claude session. Use for cross-agent handoff, not an ordinary follow-up in an existing Claude session.