common-use-claude-resource

Exhaustive adversarial review for any code change in this project — PR, patch, notebook, script, pipeline. Hunts real bugs and hidden contract violations through multi-phase, multi-subagent investigation. Use when correctness matters more than speed.

Exhaustive review of a SINGLE Python script BEFORE it is executed. Mandatory for any new or modified .py file in this project. Runs multi-subagent adversarial review and writes code_review.md adjacent to the script. Use BEFORE any `python <script>.py` invocation.

Run an ultrathink adversarial post-implementation review of the current branch, PR, or worktree diff. Use manually after multi-agent or worktree development is complete and before merging.

General-purpose parallel development workflow. Use after the user has described a coding, refactoring, testing, bug-fix, or implementation task and wants Claude Code to coordinate multiple specialized subagents or agent-team teammates with worktree isolation, tests, and adversarial review.