Skills

• SkillUSD 2

  Feature Flag Rollout

  Plans a feature's release behind a flag properly: staged exposure, the metrics that gate each stage, kill-switch behaviour verified — and the flag's own funeral scheduled so it does not haunt the codebase forever.

  Best for: Releases too risky for all-at-once

• SkillUSD 2

  API Versioning

  Plans a breaking API change without breaking your consumers: the compatibility analysis, the versioning approach fitted to your situation, the migration window — and the deprecation that actually ends.

  Best for: APIs that must change shape under live consumers

• SkillUSD 2

  API Conventions Review

  Reviews your API against the conventions that make APIs guessable — naming, status codes, error shapes, pagination, time formats — and delivers the fix list that turns three styles into one.

  Best for: APIs grown by many hands into many dialects

• SkillUSD 2

  Schema Review

  Reviews a proposed database schema before it hardens into regret: types, keys, constraints, naming, the classic traps — money as floats, naive timestamps, missing uniqueness — caught while they are still cheap.

  Best for: A schema design about to become a migration

• SkillUSD 2

  Migration Runbook

  Writes the runbook for a risky operational change — a data migration, a cutover, a major upgrade — as a minute-by-minute script with verification gates, abort criteria and the rollback rehearsed on paper first.

  Best for: The Saturday-night change that must not become a Sunday incident

• SkillUSD 2

  Test Plan

  Writes the test plan for a change before anyone tests: what must be verified, what could regress, which cases cover it at which level — so testing becomes a checklist executed, not an improvisation remembered differently by everyone.

  Best for: Changes big enough that 'I clicked around' is not evidence

• SkillUSD 2

  Test Cases from Spec

  Turns a specification into concrete test cases before (or while) the code is written: the happy paths, the boundaries, the refusals and the ugly edges — each as input, action and expected result.

  Best for: Making acceptance criteria actually checkable

• SkillUSD 2

  End-to-End Scenarios

  Chooses and scripts the few end-to-end journeys worth their cost: the flows that prove the system hangs together, written as robust scenarios — instead of an e2e suite that retests every unit test through a browser.

  Best for: Deciding what deserves end-to-end coverage (and what does not)

• SkillUSD 2

  Load Test Plan

  Designs load tests that answer real questions — can we survive the launch, where do we break first, what degrades before what fails — with realistic traffic shapes instead of a meaningless hammer at one endpoint.

  Best for: The launch or campaign you suspect will triple your traffic

• SkillUSD 2

  Performance Budget

  Sets the performance budgets your product will actually keep: concrete limits per page and operation, measured where users are, enforced where code changes — so speed is a maintained feature instead of a quarterly apology.

  Best for: Products that get slower every sprint, nobody deciding it

Also see the Agents.