Claude Skills by trungdo9

Git operations with conventional commits (cm, cp, pr, merge, finish) — scoped commits from the claim registry, finish-branch protocol with draft-default PRs + paste-ready auth fallback, and declared delivery-tail execution semantics

Build a semantic knowledge graph from unstructured text — design docs, RFCs, ADRs, transcripts — extracting entities and relations for queryable, RAG-friendly knowledge.

Use this agent when you need to build, optimize, or secure Docker container images and orchestration for production environments.

Build interactive documentation sites with Mintlify, MDX components, and OpenAPI playground integration

Implement payment integration with SePay (Vietnamese payment gateway with VietQR, NAPAS, bank transfers, and cards). Use when integrating Vietnamese payment processing, implementing checkout flows, handling webhooks, processing bank transfers, generating QR codes, or building billing systems for the Vietnamese market. Supports API Key/OAuth2 authentication, virtual accounts, bank account monitoring, and SDK integrations (Node.js, PHP, Laravel).

Use when you need to plan technical solutions that are scalable, secure, and maintainable.

Apply Kanban methodology to the plans/ folder — surface what's in progress, what's blocked, and what's next, using folder structure and frontmatter rather than external boards.

Port and refactor a feature from a public GitHub repo into the local codebase. Use when user wants to copy/adapt an existing implementation (auth flow, payment integration, UI component, algorithm, etc.) from another repo rather than build from scratch. Triggers on `/ck:port` or phrases like "port from", "borrow this feature", "copy from repo X".

Create presentations and slide decks (Marp, reveal.js, Quarto) from markdown or notebooks

Apply systematic problem-solving techniques for complexity spirals (simplification cascades), innovation blocks (collision-zone thinking), recurring patterns (meta-pattern recognition), assumption constraints (inversion exercise), scale uncertainty (scale game), and dispatch when stuck. Techniques derived from Microsoft Amplifier project patterns adapted for immediate application.

Organize directory layout, naming, and module boundaries in an existing or growing codebase — monorepo vs polyrepo, src-vs-flat layouts, conventions that scale past 1000 files.

Use when the user needs a LARGE MECHANICAL change across the codebase — rename a symbol everywhere, extract a module, migrate an API surface, run a codemod, convert patterns (e.g. class→hooks, callbacks→async/await). Distinct from `/ck:cook` (feature work) and `/ck:fix` (issue resolution). Triggers on "/ck:refactor", "rename across repo", "migrate all X to Y", "extract module", "codemod". Enforces atomic commits, dry-run, type-check gates, rollback plan.

Use when you need to research, analyze, and plan technical solutions that are scalable, secure, and maintainable.

Facilitate team retrospectives — structured backward-looking reflection on completed sprints, features, or incidents to extract lessons and drive continuous improvement.

Durable per-plan run ledger (STATE.md) so a killed multi-phase run resumes with zero re-implementation. Use for any multi-phase run (/ck:cook, /ck:flow, /ck:fix, /ck:refactor), when resuming an interrupted run, after a spend/session kill, or after context compaction. Covers the append-only event contract, the resume protocol (re-derive truth from git + gates, never trust the plan's status claims), and parallel-session safety.

Design real-world test scenarios — user journeys, role-based flows, recovery paths — independent of execution tool (pytest, Playwright, Postman, etc.).

Complete security suite — expert mindset + active code scanner. Covers OWASP 2025, supply chain, L1-L4 data flow analysis, 21 vulnerability rules with a cost-tiered default pass, TINY/SMALL/LARGE scan modes, bilingual output (vi/en). Use when scanning code for security vulnerabilities, doing security audits, or applying security mindset. Triggers on "scan security", "kiểm tra bảo mật", "security audit", "review security".

Create visually impressive HTML showcases using Tailwind CSS for product demos, landing pages, and social media

Red-green test-driven discipline — NO PRODUCTION CODE WITHOUT A FAILING TEST FIRST. Use for bug fixes (default), regression-prone changes, or when asked to "write tests first" / "TDD this". Covers red → verify-red → green → verify-green → refactor, the baseline rule (baseline-first, never git stash), the rationalization table, and red flags. Discipline only — test infra and browser toolkits live in test-automation, case derivation in scenario.

Orchestrate parallel multi-session collaboration with independent Claude Code teammates. CK-native (v2.1.0) engine that spawns independent Claude Code sessions as teammates — each with own context window, task ownership, and persistent cross-session memory. Use when 3+ independent workstreams must run in parallel, workers need to discuss findings, competing hypotheses need evaluation, or multi-layer work (API + UI + DB) benefits from per-layer ownership.

Create publication-ready SVG technical diagrams

Break a plan, spec, or the current conversation into vertical tracer-bullet tickets, each declaring the tickets that block it — written as one file per ticket under plans/<plan-dir>/tickets/, or published as issue-tracker sub-tasks under a parent issue on explicit opt-in. Use when a plan needs slicing into independently shippable units, when work is handed to parallel agents or devs, or when a wide refactor needs an expand–contract sequence. Triggers on "break this into tickets", "slice the p...

Falsify a plan's factual claims before implementing against them. Use before executing any plan written in another session (/ck:cook --from-plan), via /ck:plan verify <path>, or whenever a plan asserts existing-behaviour facts (a root cause, "X currently does Y", affected row counts, "already done"). Produces a claim → verdict → evidence table; no code until it is approved.