Best agent skills for production readiness
Cover the security, performance, observability, automation, and rollback work after code functions.
How to choose
Addy offers the explicit path from working code to an operable release. Matt’s wizard remains useful for the manual setup and cutover steps agents cannot complete themselves, while Superpowers and Pstack supply the verification evidence that should gate the release in the first place.
Wizard
Creates an interactive script for manual infrastructure, credentials, and cutover steps.
Verification Before Completion
Requires fresh evidence from real commands before claiming that work is fixed, passing, or complete.
Our take: The counterweight to optimistic agents: it requires fresh command output as evidence before any claim that work is fixed, passing, or complete. Pairing it with a planning or implementation skill removes most false success reports.
Finishing a Development Branch
Verifies tests, detects the environment, and presents a merge, pull request, or cleanup decision for completed work.
Create Verification Skill
Generates a project-local verify skill with a feature map for any language or platform.
Show Me Your Work
Keeps a reviewable decision trail in a commit-friendly TSV.
Security and Hardening
Hardens authentication, input handling, storage, integrations, and privacy boundaries.
Our take: Security and Hardening widens an agent’s implementation lens to trust boundaries, untrusted input, sessions, data handling, external integrations, and privacy. It is one of the clearest capabilities Addy’s lifecycle adds beyond an inner-loop coding methodology.
Performance Optimization
Profiles and improves frontend, backend, query, and database performance.
Observability and Instrumentation
Adds logs, metrics, traces, and alerts that make production behavior diagnosable.
CI/CD and Automation
Builds deployment pipelines with automated quality gates and release strategies.
Shipping and Launch
Prepares monitoring, staged rollout, rollback, and pre-launch verification.
Our take: Shipping and Launch makes deployment an engineering phase rather than the moment after tests pass. It forces explicit monitoring, staged exposure, rollback, ownership, and go/no-go evidence.