Orchestrate End-to-End Feature Development

A 12-step, 4-phase orchestration skill that coordinates specialized sub-agents through the full feature development lifecycle, from requirements to production deployment. Phase 1 (Discovery and Requirements) calls a business-analyst to define user stories and success metrics, an architect-review agent to design technical architecture and API contracts, and a security-auditor to assess risk and compliance. Phase 2 (Implementation) calls a backend-architect to build APIs with resilience patterns and feature flags, a frontend-developer to build UI with feature-flag-driven A/B testing, and a data-engineer to build ETL pipelines and analytics events. Phase 3 (Testing and QA) calls a test-automator for unit, integration, E2E, and performance tests targeting a minimum 80% coverage, the security-auditor again for OWASP checks and penetration testing, and a performance-engineer for profiling and query/caching optimization. Phase 4 (Deployment and Monitoring) calls a deployment-engineer to build CI/CD with feature flags, blue-green rollout, and a rollback runbook, an observability-engineer to set up tracing, dashboards, and SLOs/SLIs, and a docs-architect to generate API docs, runbooks, and architecture documentation. Each phase's output feeds as explicit context into the next agent's prompt. The rollback strategy is staged by speed: disable the feature flag in under a minute, switch blue-green traffic in under five minutes, or fall back to a full CI/CD deployment rollback in under fifteen minutes, coordinating a database migration rollback with the data team if needed before an incident post-mortem.

Use for coordinating end-to-end feature delivery that spans backend, frontend, and data, needs multi-team alignment on scope and risk, and will require a real deployment and monitoring plan. Do not use for a small, isolated backend change or bug fix, when only a single specialist task is needed rather than a full workflow, or when there is no deployment or cross-team coordination involved.

Required parameters: --feature (name and description), --methodology (traditional, tdd, bdd, or ddd), --complexity (simple 1-2 days, medium 3-5 days, complex 1-2 weeks, or epic 2+ weeks). Optional parameters configure --deployment-strategy (direct, canary, feature-flag, blue-green, or a-b-test), --test-coverage-min (default 80%), --performance-budget, --rollout-percentage (default 5%), and named tool integrations for --feature-flag-service, --analytics-platform, and --monitoring-stack. Output across the 12 steps is a requirements document, architecture document, security assessment, working backend, frontend, and data-pipeline code, a full test suite, a CI/CD deployment pipeline with rollback procedures, monitoring dashboards and SLOs, and complete documentation.

Delegates to named specialist sub-agents via the Task tool: business-analyst, architect-review, security-auditor, backend-architect, frontend-developer, data-engineer, test-automator, performance-engineer, deployment-engineer, observability-engineer, and docs-architect. Supports pluggable feature-flag providers (LaunchDarkly, Split, Unleash, or custom), analytics platforms (Segment, Amplitude, Mixpanel, or custom), and monitoring stacks (Datadog, New Relic, Grafana, or custom).

Engineering leads or platform teams orchestrating a non-trivial feature across multiple specialties who want a repeatable phase structure - discovery, implementation, testing, deployment - with explicit rollback and observability built in, rather than ad-hoc coordination.