Engineering

Creates AI-generated content policies with disclosure requirements, quality standards, and review processes. Use when standardizing how your business produces AI-assisted content.

Creates AI usage policies for businesses with transparency commitments, bias mitigation, and disclosure guidelines. Use when establishing responsible AI practices.

Identifies AI automation opportunities in business workflows with feasibility assessment and ROI estimates. Use when evaluating where AI can save time and money in your business.

Designs automation workflows with trigger-action sequences, tool connections, and error handling procedures. Use when automating repetitive business processes.

Use when cleaning up commit history before a PR. Interactive rebase, fixups, squash decisions, and how to recover when rebase goes sideways.

Use when reviewing a diff, pull request, or commit before merge. Focuses on correctness, hidden footguns, and what a senior engineer would actually flag — not nit-pickery.

Use when inheriting a codebase you didn't write. Builds a mental model fast, identifies the load-bearing parts, and pulls the right questions out of the original author before they leave.

Use when you have more tech debt than you can fix. Helps separate "biting now" from "theoretical", builds a cost/risk frame, and writes the case for leadership.

Use when migrating a JavaScript codebase to TypeScript incrementally. What to type first, the strictness escalation path, and how to handle libraries with bad types.

Use when designing or critiquing a REST, GraphQL, or RPC API. Focuses on resource modeling, versioning, errors, idempotency, pagination — the things that make an API pleasant or painful 6 months in.

Use when writing tests for new code or characterizing legacy code before refactoring. Writes tests that exercise behavior, not implementation — tests that survive refactors.

Use when there's a bug but the cause isn't obvious. Systematically isolates the failure, generates hypotheses, designs the minimal test to falsify each one. Avoids the "try random fixes" trap.

Use when the app feels slow and the cause isn't obvious. Profiles before optimizing, picks the right tool, reads flame graphs, and knows when "fast enough" is the correct answer.

Use when setting up or fixing CI/CD. Covers caching, parallelization, secret management, deploy gates, rollback, and the discipline of treating pipelines as code.

Use when designing a new schema or evolving an existing one. Covers normalization trade-offs, indexing, soft vs hard deletes, naming, and the migration mistakes that cause Sunday-morning incidents.

Use when planning a major version upgrade — Next.js, React, Node, Django, Rails, framework X. Sequences the migration, reads the CHANGELOG with intent, plans the regression pass.

Use when wiring logs, metrics, traces for a service. What to instrument, what to skip, when to alert vs log, and how to keep on-call from hating you.

Use when about to refactor working code. Enforces "characterize first, then change", uses tests as guardrails, and knows when NOT to refactor at all.

Use for security review of code, architecture, or a diff. Focuses on practical OWASP-style issues, auth/authz gaps, and secrets — not checklist theater. Read-only by design.