jwilger/agent-skills
3 skills3 installs9 starsGitHub
Install
npx skills add https://github.com/jwilger/agent-skillsSkills in this repo
1Domain ModelingDomain Modeling is an agent skill that teaches how to replace raw strings, ints, and booleans with domain-meaningful types so the compiler enforces business rules. Solo and indie builders invoke it when they want a focused review for primitive obsession, semantic types, value objects, newtype wrappers, swappable parameters, or bool-as-state—and when they are in a TDD domain review step rather than a generic lint pass. The skill frames its value as communication: code speaks the business language and turns implicit assumptions into explicit, compiler-checked contracts that humans and agents can reason about. It is independently useful for design conversations and code review, and it deliberately does not cover general formatting reviews, debugging workflows, or patterns unrelated to domain type modeling. High-effort, standalone usage with no bundled skill prerequisites makes it a drop-in quality layer on existing repos.1installs2Event ModelingEvent-modeling is an agent skill package that teaches how to write GWT (Given/When/Then) scenarios as acceptance criteria for vertical slices in event-modeling workflows. It is aimed at solo and indie builders who are breaking a domain into slices and need unambiguous “done” definitions before coding. Use it after workflow or event-stream design is complete—not as a substitute for discovering commands and events. The skill supplies concrete markdown templates for command scenarios (state change), view/projection scenarios, automation triggers, and error paths where no events should be published. That structure helps agents and humans keep acceptance language consistent with event naming and payloads. It matters because vague slice boundaries cause rework; GWT scenarios become the contract between product intent and implementation. Complexity is intermediate: you should already understand events, commands, and projections.1installs3TddTDD is an agent skill that forces test-driven development inside Claude Code and compatible agents by choosing an execution strategy up front—subagent orchestration when the Agent tool exists, otherwise chained steps in SKILL.md. The harness supplement spells out permissions subagents need, why delegate mode breaks file edits, and how to resume stopped agents with intact context rather than re-supplying delegation payloads. It also describes creating blocking RED-phase tasks so workflow state stays visible across a cycle. Solo builders install it when they want their coding agent to honor red-green-refactor instead of jumping straight to implementation, especially on backend services, CLIs, or APIs where regressions are costly. The skill is intermediate to advanced because it assumes comfort with tests, repo edits, and multi-step agent coordination. It pairs naturally with build-phase feature work and ship-phase quality gates without replacing a human judgment on what to test first.1installs