Context Survey
Use when the user asks to inspect, research, summarize, inventory, compare, understand existing code/docs, gather context, validate claims, or answer what is true before planning or changing files.
Implementation
Use when the user asks to implement, add, change, edit, update, wire, migrate, fix after diagnosis, or execute approved scoped code/content/config/documentation changes.
Project Audit
Use when the user asks for a project audit, health check, repo/tooling/package/skill health scan, maintenance review, architecture smell review, drift detection, or risk assessment.
Shipping
Use only when the user explicitly asks to ship, commit, prepare a PR, merge, tag, publish, release, deploy, package, hand off, or finalize already-completed work.
Refactoring
Use when the user asks to refactor, clean up, simplify, extract, inline, reduce duplication, improve maintainability, address code smells, improve boundaries/types, or reorganize code without intended behavior changes.
Root Cause Analysis
Use when the user reports a bug, regression, failing test, error, flaky behavior, unexpected output, broken workflow, suspicious logs, performance anomaly, or asks for RCA/root cause.
Task Creation
Use when the user wants implementation steps, task breakdown, tickets, issues, milestones, phases, vertical slices, sequencing, or agent-ready work from an approved goal or spec.
Product Planning
Use when the user wants brainstorming, product planning, UX or copy shaping, architecture direction, scope decisions, acceptance criteria, or a spec after details are finalized.
Change Review
Use when the user asks for code review, change review, diff/branch/PR review, regression review, readiness assessment, critique, or blocker finding without edits.