Safe Launch
Guides engineering teams through observable, reversible, and incremental production releases.
The skill emphasizes monitoring, staged rollout, feature flags, and rollback, which reduce external impact. However, it does not require user confirmation before deployment, pushing, or database rollback, and it lacks explicit privilege boundaries, data-flow disclosure, sensitive-log handling, and pre-rollback backup guidance, so points are deducted.
The document is internally coherent and provides checklists, thresholds, and failure paths. However, its examples depend on undeclared npm, Prisma, React, monitoring, and deployment environments, with no executable tests for this workflow or detailed diagnostic feedback; static calibration therefore limits the score.
The triggers and core scenarios are clear, covering production deployment, migrations, beta releases, canaries, and rollback. Boundaries, required inputs, alternative stacks, and Chinese-user or mainland-China environment considerations are not specified, and several recommendations assume a JavaScript/Web stack, so points are deducted.
The skill has clear layering with overview, triggers, process material, examples, anti-rationalization guidance, red flags, and verification. Repository-level MIT licensing, CI, and maintainer information exist, but the skill lacks independent versioning, changelog, owner, update path, and dependency-installation notes, so points are deducted.
The checklist, rollout sequence, monitoring metrics, thresholds, and rollback template are directly reusable for core launch planning. Yet thresholds lack cited context, the database rollback example may not fit the target system, and substantial project-specific adaptation remains necessary, so full marks are not justified.
The skill contains explicit checks, thresholds, and verification steps, and repository CI validates skill content. However, the CI does not cover this skill's key launch paths, the supplied test is unrelated, and there is no third-party attribution or reproducible execution evidence; evidence is therefore limited under static review.
- Deployment, git push, and database rollback examples may cause irreversible or high-impact external effects; confirm target environment, permissions, backups, and rollback conditions first.
- The thresholds and commands are generic examples and should not be adopted without project validation; add stack-specific tooling, migration strategy, and diagnostic failure guidance.
- No skill-specific tests or third-party support are provided, so static review cannot establish that the launch workflow is executable in practice.
What does this skill do, and when should you use it?
Shipping and Launch is an Agent Skill for preparing production deployments, covering pre-launch checks, feature flags, staged rollouts, monitoring, and rollback. It treats every launch as reversible, observable, and incremental, using staging, internal users, a 5% canary, and progressive expansion to reduce risk. Its checklist spans code quality, security, performance, accessibility, infrastructure, and documentation. It suits teams that need explicit release gates and incident responses, but it does not execute a real deployment or provide cloud-specific configuration.
It guides an agent through checks for tests, builds, linting, type checking, review, vulnerabilities, authentication and authorization, performance, WCAG 2.1 AA, production configuration, database migrations, DNS, SSL, CDN, logging, error reporting, and health checks. It defines a feature-flag lifecycle covering deployment, enablement, rollout, monitoring, and cleanup, and lays out a 5% → 25% → 50% → 100% rollout sequence. It supplies advance, hold, and rollback thresholds for error rate, P95 latency, client JavaScript errors, and business metrics. It also requires documented rollback triggers, steps, database considerations, estimated recovery times, and first-hour post-launch verification.
- A frontend or backend engineer deploying a new feature to production for the first time needs a release-readiness checklist.
- A technical lead releasing a major or high-risk change needs feature flags and canary rollout controls.
- A team migrating a database or infrastructure needs to plan migration rollback and data handling in advance.
- A product engineering team opening a beta or early-access program needs to monitor errors, latency, user behavior, and business metrics.
- An engineering team establishing production monitoring, error reporting, and health-check procedures needs a structured launch workflow.
What are this skill's strengths and limitations?
- Covers release gates across code, security, performance, accessibility, infrastructure, and documentation.
- Provides concrete feature-flag lifecycle, rollout stages, and rollback thresholds.
- Includes first-hour post-launch verification, error reporting, and database rollback considerations.
- Emphasizes evidence-based verification, monitoring, and reversible releases.
- It does not replace documentation for a specific cloud platform, deployment system, or monitoring product.
- It provides no directly executable deployment scripts or platform-specific integrations.
- Its thresholds need adjustment to the service's baselines, business metrics, and risk profile.
- The source provides no test suite, execution results, or cross-client empirical compatibility evidence.
How do you install this skill?
The README documents installation for the full 24-skill collection: run npx skills add addyosmani/agent-skills, or run npx skills add addyosmani/agent-skills --list to browse first. It also documents the CLI pattern for individual skills; npx skills add addyosmani/agent-skills --skill shipping-and-launch is the corresponding command, although the README does not explicitly show this exact skill name. For Codex, run codex plugin marketplace add addyosmani/agent-skills; after installation, skills are read from the repository's skills/ directory. The source does not fully document per-client folder placement for this individual skill.
How do you use this skill?
Trigger shipping-and-launch when preparing a production deployment, releasing a significant change, migrating data or infrastructure, opening a beta, or needing a pre-launch checklist, monitoring plan, staged rollout, or rollback strategy. A suitable prompt is: “Prepare and walk through the pre-launch checklist, staged rollout plan, and rollback plan for this production release.” The source defines no dedicated trigger command or executable automation script; the skill provides the workflow, thresholds, and verification checklist for an agent to follow.
How does this skill compare with similar options?
The README explicitly names Superpowers and Matt Pocock's skills and links to a comparison document, but the supplied source does not include the comparison details, so no further judgment is made.