Reusable SKILL.md capabilities that give your agent new expertise — installed into .claude/skills/ and friends.
.claude/skills/
Adversarial code review using the opposite model. Spawns 1–3 reviewers on the opposing model (Claude spawns Codex, Codex spawns Claude) to challenge work from distinct critical lenses. Triggers: "adversarial review".
Negative perspective code review (convention, quality, bugs, performance)
Thoroughly address ALL PR comments using parallel Opus subagents. Each subagent invokes /task skill.
Perform thorough Linus-style code review focusing on correctness, maintainability, and adherence to project conventions. Use after completing implementation to ensure code quality. Invoked via /sdd-review [file-path or PR-number].
Architect Alphonso conducts rigorous code review and architecture-fit analysis: ADR compliance, test coverage, architectural patterns, security. Outputs review document with APPROVED/REDIRECT/BLOCKED status.
Use the @steipete/oracle CLI to bundle a prompt plus the right files and get a second-model review (API or browser) for debugging, refactors, design checks, or cross-validation.
Launch lgtm TUI to review a markdown file with line-by-line commenting