apps-in-toss-harness

In the author's own words, from each skill's SKILL.md.

• debug

  Debug an Apps in Toss mini-app across two environments — local browser (devtools panel, `window.__ait`, browser DevTools) and on-device intoss-private candidate (real Toss WebView via `ait-devtools` MCP relay-staging QR attach). Branches by what it observes, then runs an autonomous observe → hypothesize → minimal-fix → re-verify loop and hands off when it stalls (§6). Triggered by `/ait:debug` (n…

  packages/agent-plugin/shared/skills/debug/SKILL.md on GitHub (opens in a new tab)

• design

  Build and fix mini-app screens. Run it for "화면 만들어줘", "출석 체크 화면 그려줘", "화면이 구린데 예쁘게 해줘", "디자인 좀 봐줘" — it writes and edits the project's actual screen files, not a chat answer about design. Works from nothing: with no screen, no CSS file and no design assets in the project, it still produces the screens. On projects built with TDS, MUI, Tailwind, shadcn or their own tokens it keeps that system, so…

  packages/agent-plugin/shared/skills/design/SKILL.md on GitHub (opens in a new tab)

• inject

  Patch an existing Apps in Toss mini-app's build setup. Three facets: `/ait:inject-devtools` adds the `@apps-in-toss/devtools` unplugin for browser dev ("기존 Vite 프로젝트에 devtools 붙여줘"); `/ait:inject-debug-console` installs `@apps-in-toss/debug-console` (on-device attach) as a dependency; `/ait:inject-tossface` wires the Tossface emoji web font (CDN link or subset bundling) into a project's entry CSS…

  packages/agent-plugin/shared/skills/inject/SKILL.md on GitHub (opens in a new tab)

• launch

  Pre-launch check, review request and rejection handling for an Apps in Toss mini-app (검수 = the Apps in Toss app review, not code review). Use it right away for "검수 넣기 전에 한번 봐줘", "출시해도 되는지 점검해줘", "검수 요청 넣어 줘", or a pasted rejection ("검수에서 반려됐어. 사유가 …"); it reads the project itself, so don't ask which files first. Checks code against the official checklist and console review state, and drafts an in…

  packages/agent-plugin/shared/skills/launch/SKILL.md on GitHub (opens in a new tab)

• new-miniapp

  Scaffold a new Apps in Toss mini-app by driving the official `create-ait-app` CLI non-interactively (`--inline`, always `@latest`), then verifying the `@apps-in-toss/devtools` wiring (mock SDK + panel) the CLI performs so `npm run dev` runs in a plain browser immediately — falling back to manual wiring only when the CLI did not do it. Supports `--tds` and `--sample iap,iaa` passthrough. Falls bac…

  packages/agent-plugin/shared/skills/new-miniapp/SKILL.md on GitHub (opens in a new tab)

• plan

  Take a mini-app from a vague idea to a plan — ideate when the user has no concrete idea yet, shape a lightweight PRD, look up official docs for reference, then map it to the needed SDK domains, the runtime permissions they prompt for, and console terms (약관) that gate registration. Use when the user asks "미니앱 만들고 싶은데 뭘 만들지 모르겠어" or "미니앱 만들 건데 필요한 SDK 도메인/권한/약관 먼저 정리해줘", or asks for a PRD ("출시 일정이랑…

  packages/agent-plugin/shared/skills/plan/SKILL.md on GitHub (opens in a new tab)

• server

  Write or review the server side of an Apps in Toss mini-app: any backend that stores per-user data or calls the Apps in Toss partner API (Toss login token exchange, anonymous-key exchange, other partner calls), on Cloudflare Workers, Node, a BaaS (Supabase, Firebase) or any other hosting. Use it right away, before writing server code, for "서버에 저장해서 다른 기기에서도 보이게 해줘", "서버 붙여줘", "Supabase·Firebase에…

  packages/agent-plugin/shared/skills/server/SKILL.md on GitHub (opens in a new tab)

• setup-debugger

  Wire the `ait-devtools` debug MCP server (`@apps-in-toss/debugger`) into the current project's MCP config (project scope, opt-in) — `.mcp.json` on Claude Code, `.cursor/mcp.json` on Cursor — so `/ait:debug` can run on-device CDP relay attach (environment 3). Idempotent JSON merge — preserves other server entries, never renames the `ait-devtools` key. Run THIS skill instead of hand-writing `.mcp.j…

  packages/agent-plugin/shared/skills/setup-debugger/SKILL.md on GitHub (opens in a new tab)

• test-on-device

  Put the current mini-app on a real phone through the standard path — build the `.ait` bundle, upload it with console MCP (`bundle_upload` → `bundle_upload_complete`), confirm the compile with `miniapp_get_status`, and hand over the entry link the tools returned so it opens in the Toss app. Review submission, release, rollback, and promotion are out of scope. Triggered by `/ait:test-on-device`, no…

  packages/agent-plugin/shared/skills/test-on-device/SKILL.md on GitHub (opens in a new tab)

• ux-writing

  Rewrite mini-app screen copy against measurable UX writing principles. Use when the user asks "문구 다듬어줘", "카피 점검해줘", or "UX 라이팅 기준으로 봐줘" — invoke it right away even with no screen or file named yet, it surveys the current project's screen copy first, so do not ask a clarifying question or answer with generic advice instead of running it. This is a structured rewrite pass, not a one-off inline answ…

  packages/agent-plugin/shared/skills/ux-writing/SKILL.md on GitHub (opens in a new tab)

• welcome

  Print the harness entry-point map right after plugin install, and check the local environment/integration state (git, Node/npm/npx, cwd shape, MCP wiring) to suggest or hand off the next concrete step — since `/plugin install` itself prints no next step. Station-0→1 hand-off. Triggered by `/ait:welcome`, no args.

  packages/agent-plugin/shared/skills/welcome/SKILL.md on GitHub (opens in a new tab)

No skill matches that filter.

Similar skills

• reaby morluto Reverse engineer anything with agents, from app behavior down to native binaries. 366k stars · 1 skill
• superpowersby obra An agentic skills framework & software development methodology that works. 304k stars · 15 skills
• skillsby mattpocock Skills for Real Engineers. Straight from my .agents directory. 301k stars · 30 skills
• ECCby affaan-m The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Cod… 287k stars · 30 skills
• andrej-karpathy-skillsby multica-ai A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls. 223k stars · 1 skill
• skillsby anthropics Public repository for Agent Skills 183k stars · 19 skills