Skill Claude CodeCodex
Use when the user asks for adversarial, independent, skeptical, panel, multi-agent, second-opinion, high-scrutiny, security-minded, or critical review of a PR, branch diff, uncommitted changes, changed files, plan, or specific paths.
Skill Claude CodeCodex
Use when reviewing a diff, pull request, or changed files for caller-facing interface quality across APIs, exported functions, service methods, hooks, component props, schemas, commands, events, or extension points. Focus on boundary shape rather than general correctness.
Skill Claude CodeCodex
Use when leading a larger implementation, investigation, audit, migration, or review that needs an observable end state, explicit scope, a stopping condition, bounded parallel work, or evidence matched to completion claims. Do not use for a small task with one obvious check.
Skill Claude CodeCodex
Use when writing or reviewing an implementation plan, scoping a refactor, designing a feature, choosing between architectural approaches, or turning a broad change request into a decision-complete sequence of independently verifiable steps. Do not use for active root-cause investigation.
Skill Claude CodeCodex
Use when creating, migrating, running, reviewing, or debugging repository-owned recurring, scheduled, webhook, event-driven, or agent-run automations; deciding whether behavior belongs in code, a playbook, or an external scheduler; or establishing a lightweight convention for automation instructions. Routes work to…
Skill Claude CodeCodex
Use when refining recently written code, reviewing a diff for unnecessary complexity, doing cleanup passes, removing dead indirection, collapsing duplicate concepts, reducing helper bloat, or checking whether a change can be smaller without losing clarity.
Skill Claude CodeCodex
Use when behavior is broken, regressed, flaky, inconsistent, unexpectedly slow, or failing in tests, local development, CI, staging, or production and the root cause is not yet proven. Use for diagnosis before proposing or implementing a fix.
Skill Claude CodeCodex
Use when onboarding to an unfamiliar repository, module, service, feature, or execution path; answering how a subsystem works; mapping architecture before a change; or identifying the exact entrypoints, boundaries, contracts, and files to read first.
Skill Claude CodeCodex
Use when creating, editing, packaging, consolidating, routing, validating, or publishing agent skills under .agents/skills or a tool-specific skill directory.