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 |
| Apply Lean to software delivery | lean-sensei | Improve flow, adapt coaching, encourage design discovery |
| Build with hacker ethos | hacker-manifesto | Solve, share, automate drudgery, earn competence |
| Readability and CUPID review | clean-code | Human-scale design, naming, complexity, comments, CUPID |
| SOLID, if/else/switch, objects vs data | uncle-bob-design-principles | Uncle Bob: SOLID relevance, one-decision factories, class vs data-structure trade-offs |
| Scope, typed values, naming | general-programming-habits | Effective Java ch. 9: local scope, strings vs types, code to interfaces, naming |
| Review the diff | review | Review current diff, evidence, risk, and approved fixes |
| Mine review feedback | fetch-pr-corpus | Fetch merged-PR review corpus as JSONL for style mining |
| Show changes as diffs | show-me-the-code | Show unified diffs when explicitly requested |
| Ship with human ownership | code-ownership | Confirm intent, validation, and approval before shipping |
| Root-cause a defect | dantotsu | Contain defects, investigate causes, verify prevention measures |
| Test the right things (+ TCR, smoke) | testing | Test portfolio, flakiness, mocks, BDD, AI |
| Clarify a BDD story | example-mapping | Map rules, examples, questions, and scope before development |
| Write smoke tests | smoke-tests | Identify + gate CI on critical-path validation |
| TDD rigorously | tcrdd | Test-first loop; on-demand git-cycle, autonomy and exception references |
| Refactor cleanly | clean-code → refactoring | Tidy just in time; 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 |
| Architecture boundaries | the-architect | Choose clean, hexagonal, or functional-core boundaries |
| JavaScript & TypeScript mastery | typescript, type-system, zod, composition, Deno guidance | Strict types, JS idioms, runtime validation, Deno; FP techniques loaded on demand |
| Lisp, Clojure & CHICKEN Scheme | lisper | Lisp dialects, Clojure/CLJS idioms, CHICKEN tooling, interop |
| Idiomatic Go | golang, golang-testing, plus sub-skills | Project layout, error-handling, concurrency, generics, testing, Go 1.24–1.27 |
| Idiomatic Ruby / Rails | ruby | Style, GoF patterns, RSpec/Minitest testing |
| Verify PHP against php.net | php | Check signatures, edge returns, PHP 8 removals |
| OCaml | ocaml | OCaml/dune; extracted guidance on invariants, collections, memoization, and graphs |
| Expert C/C++ | c-cpp | Value semantics, pointers/references, C foundations and interoperability |
| ReScript | rescript | Write/debug ReScript, JSX v4, JS interop |
| Git as a craft | git-hero, plus sub-skills | Git help (FR / EN), atomic commits, Gitmoji, GitLab DAG |
| Security scan (Trivy + ggshield) | security-scan | Vulns, misconfig, leaked secrets; severity-ranked report |
| REST API design | rest-api-design | Core API decisions; on-demand response, collection and reliability references |
| System diagrams | c4-diagram + explain-code | C4 / Mermaid diagrams and code walkthroughs |
| Markdown authoring | markdown | GitHub-Flavored Markdown syntax reference |
| Sharpen communication | communication | Analyze artifacts; apply rhetoric and author voice presets |
| Communicate in an author voice | communication → presets | DHH, Uncle Bob, Jessitron, and Martin Fowler presets |
| Daily progress log | logging-daily-progress | Rolling commit-based impact / learnings report |
| Write README / AGENTS.md well | repo-docs | Draft/audit README, AGENTS.md, CLAUDE.md; Diátaxis doc sets |
| Ground on real library source | vendor-reference | Vendor library gitignored, read source, never commit |
| Write idiomatic systems Rust | rust | Ownership, errors, traits, concurrency, unsafe, cargo |
| Idiomatic modern Java / AspectJ | java | Effective Java themes, records, streams, concurrency, AspectJ |
| Design data-intensive systems | data-systems | Data models, replication, transactions, streams, database choice |
| Idiomatic Elixir / OTP | elixir | Pattern matching, pipelines, OTP supervision, Mix, ExUnit |
| Safe Elm front ends | elm | Elm architecture, types, decoders, ports, tooling |
| HTML-first front ends | html-first | htmx hypermedia + Lit web components |
| Live Smalltalk in Pharo | smalltalk-pharo | Image-based Pharo, message passing, Spec, MVC history |
| Idiomatic modern Python | pythonista | Idioms, typing, pytest, packaging, asyncio, web |
Install lisper for the complete Lisp package, including CHICKEN Scheme's compiler, eggs, FFI, and deployment references. Dialect-specific guidance loads on demand; no separate CHICKEN skill is required.
The TypeScript skill follows each project's configured runtime, imports, tests, and verification workflow.
Its Total TypeScript reference adds guidance from Matt Pocock's public material on type derivation, satisfies, generic inference, conditional types, and type tests.
Its JS pitfalls and idioms reference distills coercion, dictionary, iteration, API-design, and async rules from Effective JavaScript, The Good Parts, JavaScript Patterns, and You Don't Know JS.
Its ES features by edition reference summarizes ES2015–ES2025 additions from the ECMA-262 specification.
Its domain modeling reference adapts Domain Modeling Made Functional to TypeScript: workflows, Result pipelines, DTO boundaries, and persistence.
For Effect TS, the skill links to the official documentation and directs agents to match the installed library version.
See .benchmarks/README.md for instruction-uptake benchmarks and their limits.
- Create
skill-name/SKILL.mdwith frontmatter (name,description) and your core workflow - Push to a shared repo the assistant reads from
See repo-docs for quality guidance and the router pattern (nested sub-skills).
Built by: birrame · Last updated: Oct 17, 2026