templates/skills/review/arch/full-check/SKILL.md
--- name: review-arch-full-check description: Full architecture review: decisions, trade-offs, NFRs, risks, and rollback plan --- <when_to_use> <trigger>Major changes to architecture or public contracts</trigger> <trigger>High risk in reliability, security, or scalability</trigger> <trigger>Before launching an initiative with long-term consequences</trigger> </when_to_use> <input_requirements> <required>ADR or equivalent document with options and selected approach</required> <require
npx skillsauth add ozerohax/assistagents templates/skills/review/arch/full-checkInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
<when_to_use> <trigger>Major changes to architecture or public contracts</trigger> <trigger>High risk in reliability, security, or scalability</trigger> <trigger>Before launching an initiative with long-term consequences</trigger> </when_to_use>
<input_requirements> <required>ADR or equivalent document with options and selected approach</required> <required>Change goals and success criteria</required> <required>Description of impact on NFRs: reliability, performance, security, operability</required> <required>Rollout and rollback plan</required> <optional>Cost estimate and team constraints</optional> </input_requirements>
<full_review_blocks> <block>Context and problem: why the current state is insufficient</block> <block>Options and trade-offs: which alternatives were considered and why they were rejected</block> <block>Decision traceability: links to ADRs, tasks, and implementation</block> <block>NFR analysis: reliability, performance, security, observability</block> <block>Risks and failure modes: what can go wrong and how to detect it</block> <block>Evolution and reversibility: phasing, migrations, rollback</block> </full_review_blocks>
<quality_rules> <rule importance="critical">A decision without explicit trade-offs and consequences is not considered ready</rule> <rule importance="critical">Critical risks have mitigations and owners</rule> <rule importance="high">There are measurable signals of post-release success and degradation</rule> <rule importance="high">The rollout/rollback plan is realistic and verifiable</rule> </quality_rules>
<do_not> <item importance="critical">Do not approve architecture without validating NFR impact</item> <item importance="high">Do not leave irreversible changes without a rollback strategy</item> <item importance="high">Do not stop at diagrams without operational implications</item> </do_not>
<output_requirements> <requirement>Final status: approve, approve-with-conditions, or reject</requirement> <requirement>List of launch conditions and mandatory follow-up steps</requirement> <requirement>Explicit risk register with severity and owner</requirement> </output_requirements>
testing
Visual regression (baseline diffs, UI stabilization)
testing
Design detailed reproducible test cases (Given/When/Then, one check per case); not smoke/regression checklists
development
Mobile testing (native/hybrid/web, gestures, offline, interruptions)
testing
Load testing (k6/JMeter scenarios, benchmarks)