/skills
Reusable capabilities for AI coding agents, installable with a single command.
npx skills add agustinusnathaniel/skillsPlugin installation for Claude Code, Codex, Kimi, omp, ZCode, Cursor, and Devin
Install the whole collection as a plugin, or use the picker above to select individual skills.
Claude Code
/plugin marketplace add agustinusnathaniel/skills
/plugin install agustinusnathaniel-skills@agustinusnathaniel-skillsCodex CLI
codex plugin marketplace add agustinusnathaniel/skills
codex plugin add agustinusnathaniel-skills@agustinusnathaniel-skillsKimi Code CLI
/plugins install https://github.com/agustinusnathaniel/skills
/reloadomp
omp plugin marketplace add agustinusnathaniel/skills
omp plugin install --scope project agustinusnathaniel-skills@agustinusnathaniel-skillsZCode
Open a workspace, then go to Settings → Plugins → Create → Add marketplace and enter agustinusnathaniel/skills. Find it under Personal, then install agustinusnathaniel-skills.
Cursor
Copy this repository to ~/.cursor/plugins/local/agustinusnathaniel-skills, restart Cursor or run Developer: Reload Window, then find the skills under Customize.
Devin
Go to Customize → Plugins → Add plugin → From repository and enter agustinusnathaniel/skills, or run devin plugins install agustinusnathaniel/skills. Devin loads the software engineering skills only.
Software Engineering
Skills for architecture decisions, admin panels, Expo app foundations, calculations, client trust, configuration, CI workflows, hydration, URL state, TypeScript conventions, React app structure, codebase dossiers, pruning, guard triage, pull request readiness and review, testing, and tool adoption.
/admin-panel-engineering
Use when building admin panels, backoffice apps, or internal tools: adding a page, sidebar, route, or action that only some permissions or roles may see; touching login, logout, session expiry, or role edits; scoping data to a tenant, partner, or organization; or building filtered list pages, exports, audit trails, destructive confirmations, operator status messages, or approval flows.
/architecture-decision-framework
Use when comparing consequential architecture or implementation approaches, selecting technology, or revisiting a design after changed constraints or feedback.
/bulk-import-engineering
Use when building, fixing, or auditing data import and export: bulk file imports with validation previews, job status, and safe recovery; backup and restore; and JSON, CSV, or spreadsheet import and export, including in offline-first and local-first apps. Excludes ordinary single-file uploads and file analysis.
/calculation-correctness
Use when implementing or reviewing code that computes money, dates, units, counters, or derived metrics: totals, fees, interest and penalties, rounding, invoice or sequence numbers, timezone and day boundaries, unit conversion, aggregated statistics and charts, and edits that reconcile lists of rows that carry state.
/challenge-every-guard
Use when a change adds guards, retries, locks, fallbacks, or new abstractions, when a review round keeps adding mechanism, when a PR has grown far past its feature, or when reviewing a backend change to services, jobs, events, queries, or deploys. Sizes each guard to evidence and fits the change to the sibling code it lands in.
/client-trust-boundary
Use when deciding what a web or mobile client may hold and what the server must enforce: client-exposed environment variables or API keys, calling Firebase, Supabase, or Convex from the client, gating features by role in the UI, storing sessions or roles in browser or device storage, or reviewing a frontend for exposed secrets.
/codebase-dossier
Invoke by name to write an evidence-cited dossier of a codebase or several related repos: purpose, architecture, operations, history, decisions, and sharp edges.
/config-drift
Use when changing environment variables, build-time config, security headers or CSP, service worker or update settings, hosting or platform config, shared CI workflows, or agent and tool config, and when a setting has no effect or a feature silently stops working.
/expo-app-foundations
Use when building or changing the foundations of an Expo or React Native app: environments, release channels, and over-the-air updates; remote config, forced updates, or feature flags; token refresh, push notifications, and deep links; web views; or native modules, config plugins, and permissions.
/github-actions-authoring
Use when adding CI jobs, caching, or timeouts; building a reusable workflow or composite action that other repos call; automating commits, releases, or dependency-update PRs; or debugging a workflow that passes without checking anything or behaves differently than its YAML suggests.
/hydration-safe-state
Use when debugging hydration mismatches, a flash of the wrong theme, or a page that differs between navigation and reload; or when persisting state to localStorage or cookies, adding a theme toggle, restoring an auth session from storage, or showing viewer-local time or locale. React-specific notes are labelled.
/prune-codebase
Use for substantial codebase pruning, reduction, or removal of overengineering across production code, tests, documentation, agent instructions, configuration, and tooling, including follow-up passes after a reduction PR. Excludes ordinary feature work and minor cleanup.
/pull-request-readiness
Use when splitting work into PRs, building or fixing stacked PRs, deciding whether a PR is ready for review, marking it ready or requesting review, triaging bot and CI findings, or preparing the merge and deploy steps for the user.
/pull-request-review
Use when asked to review someone else's pull request, draft review or line comments, or reply in a review thread. Leads with must-fix findings, marks minor points non-blocking, stays in the user's own voice, and posts only after the exact text is approved.
/react-app-structure
Use when adding a page, route, feature, API call, or shared component to a React web app, when moving logic out of a component, or when deciding where code goes in a Vite or Next.js app. Follows the repository's existing layout first.
/test-strategy
Use when deciding what to test and at which boundary (unit, integration, E2E), reproducing a bug before patching it, planning failure-first coverage, choosing artifact-producing E2E for complex features, judging whether existing tests earn their keep, or changing model-driven behavior.
/tool-evaluation-and-adoption
Vet an open-source tool, library, package, plugin, skill, or MCP server before adopting it. Use when asked to adopt, switch to, or upgrade one, to install one, or to triage dependency and supply-chain alerts.
/typescript-code-conventions
Use when writing or changing TypeScript or TSX in any app, service, library, or CLI: function signatures, control flow, errors and async code, types, imports and exports, file names, and React components, or when a lint rule on any of these fails. Fits the repository first, these conventions second.
/url-state
Use when adding search params, filters, pagination, or sort to a URL; building shareable, bookmarkable, or deep links into lists; syncing a search box with the address bar; or when a shared link crashes, opens the wrong view, or leaks the sender's state. React-specific notes are labelled.
Automation
Recurring improvement loops that run without per-step direction.
/autonomous-improvement-loop
Compound small improvements on a recurring schedule without per-cycle direction. Use when running unattended improvement cycles over one repo or a portfolio. Learns per-target rules from merged and closed deliveries.
No skills found matching your search.