A coding-agent skill library — domain expertise packs that load on demand for any compatible assistant.
Every coding assistant is broad by default; to be sharp on a specific craft you ship a focused knowledge pack into a shared location the host reads from. When your prompt touches the right topic, the matching skill activates automatically.
Install → describe your problem → get a tailored workflow. No manual loading, no configuration.
| Why? | Skill | What it does in 7 words |
|---|---|---|
| Should I build this? | kano | Classify features by user benefit, refuse waste |
| Keep it simple | simple | Refuse complexity; code for human affordance |
| Readability pass | clean-code | Naming, complexity budgets, comment discipline |
| Review the diff | review | Read actual diff, tier feedback, apply fixes |
| Show changes as diffs | show-me-the-code | Every code change as a unified diff |
| Root-cause a defect | dantotsu | Turn one defect into permanent quality gain |
| Test the right things | testing | Tiered testing philosophy, desiderata for any lang |
| Write smoke tests | smoke-tests | Identify + gate CI on critical-path validation |
| TDD rigorously | tcrdd | Test-Commit-Revert workflows with test-first discipline |
| Fix code bloaters | refactoring | Detect long methods, large classes, primitive obsession |
| Safe incremental refactor | mikado-method | Step by dependency graph to unblock the next move |
| OOP principles review | object-oriented-programming | SOLID, GoF patterns, calisthenics, tell-don't-ask |
| CUPID review | cupid-checker | Composable, Unixy, Predictable, Idiomatic, Domain-aligned |
| Hexagonal architecture | ports-adapters-architecture | Apply the hex pattern to domain boundaries |
| Plain JavaScript rules | javascript | Name things right, handle this, avoid null traps |
| TypeScript mastery | typescript, type-system, zod | Strict types, error handling, runtime validation |
| ClojureScript | clojurescript | Write/debug/configure CLJS + JS interop |
| Functional composition | composing-software | Pipe → compose → combine techniques in JS |
| Monoid abstraction | monoids | Recognize/apply monoids in TS and CLJS |
| Idiomatic Go | golang, plus sub-skills | Project layout, error-handling, concurrency, generics, testing |
| Idiomatic Ruby / Rails | ruby | Style, GoF patterns, RSpec/Minitest testing |
| CHICKEN Scheme | chicken-scheme | Write / package / debug Scheme programs |
| OCaml | ocaml | Write/review/debug OCaml and dune projects |
| Git as a craft | git-hero, plus sub-skills | Atomic commits, gitmoji, CLI guru (FR / EN), GitLab DAG |
| REST API design | rest-api-design | URIs, verbs, pagination, errors, security, HATEOAS |
| System diagrams | c4-diagram + explain-code | C4 / Mermaid diagrams and code walkthroughs |
| Markdown authoring | markdown | GitHub-Flavored Markdown syntax reference |
| Sharpen communication | communication | Analyze and rewrite real messages/emails/transcripts |
| Daily progress log | logging-daily-progress | Rolling commit-based impact / learnings report |
| Write AGENTS.md well | writing-a-good-agents-md | Author/audit agent context files (AGENTS.md, CLAUDE.md…) |
- Create
skill-name/SKILL.mdwith frontmatter (name,description) and your core workflow - Push to a shared repo the assistant reads from
See writing-a-good-agents-md for quality guidance and the router pattern (nested sub-skills).
Built by: birrame · Last updated: Aug 23, 2026