[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77N408EESQJWW2F7CQ7G95W":3,"skill-same:01M77N408EESQJWW2F7CQ7G95W":43},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":13,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":42},"01M77N408EESQJWW2F7CQ7G95W","task",null,"Builds one change to this repo: reads the skills that govern it, plans as much as the work type needs (hotfix: none; change: printed; new: approved first), does the work and runs the gates at the plan's risk. Runs inside \u002Fflow for every request, or when the user types \u002Ftask \u003Crequest>.","therewardstore\u002Fapplaudiq-embed-cordova","therewardstore","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F229159885?v=4","01M75VGG5G8BDFNESM38GF3AB8","01M77N2WBAV78YTJBWWYVBF81E",0,"MIT","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Ftask\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: task\ndescription: >-\n  Builds one change to this repo: reads the skills that govern it, plans as much as the work\n  type needs (hotfix: none; change: printed; new: approved first), does the work and runs the\n  gates at the plan's risk. Runs inside \u002Fflow for every request, or when the user types \u002Ftask \u003Crequest>.\ndisable-model-invocation: true\n---\n\n# Task — read, plan by type, work, gate\n\nRequest: $ARGUMENTS\n\nMost bad changes start with typing before reading the rule that governs the area. `\u002Ftask` is the\norder of work that prevents that: read first, plan as much as the work type needs, build, gate. A\nskill beats a doc; the source beats both, and the doc gets fixed. Every change follows the rules in\n`CLAUDE.md` and this repo's own skills. With `TASK_MODE=auto`, `\u002Fflow` runs this for every request\nand owns the ticket, the commit question and the ship; alone, `\u002Ftask` ends at a green gate.\n\n## 0. Type, then how much stops you\n\n| Type | Plan | Gates in step 4 |\n|---|---|---|\n| `hotfix` | none: print the read-only diagnosis (cause, `file:line`, the one fix) and go | `low` |\n| `change` | print the plan and continue; stop for approval only when a step touches auth, payments, routing infrastructure or shared state | `medium` |\n| `new` | approved before the first edit — Claude Code: call EnterPlanMode now, steps 1–3 read-only, ExitPlanMode, wait; Cursor, Codex, Antigravity: the tool's plan mode (`\u002Fplan`), or print the plan and wait for `go` | `high` |\n\nA hotfix adds no dependency or migration and stays inside one module (`no-deps,no-migrations,module-scope`); one that\nmust is a `change`. Confirm the type in one line. An edit before the stop a type requires is a failed run.\n\n## 1. Read what governs the change — before the plan\n\n| The change touches | Read first |\n|---|---|\n| _(no domain skills yet)_ | — |\n| a module's logic | that module's doc (see `scripts\u002Fgit\u002Fdoc-map.txt`), and its skill if it has one |\n| a component, hook or helper, or any UI | this repo's own UI skill (listed in CLAUDE.md § Skills) and the components it already has |\n| a gate, a hook, a commit or a branch name | `docs\u002FGATES.md`; `CLAUDE.md` § Rules |\n\u003C!-- repo-routes:start -->\n\u003C!-- repo-routes:end -->\n\nIf no row matches, say so and add a **proposed skill** to the plan (name, scope, first three\nrules); it lands as its own change, never in this task. A `figma.com` URL in the request, or a\nscreen change, with the Figma MCP connected: `figma:get_design_context` on the node (and\n`get_variable_defs` for tokens) goes under Evidence; not connected → `Design: Figma MCP not\nconnected` and continue.\n\n## 2. Classify and plan — read-only\n\n- **Surfaces** — a view (`src\u002Fmodules\u002F\u003CModule>\u002Fviews`), a module store (`store\u002F`), the router,\n  the platform slice (`src\u002F{lib,config,store}`), an adapter (`src\u002Fpackages`), copy (`—`), the build.\n- **Ships in** — web.\n- **Risk** — the type sets it (`hotfix` low, `change` medium, `new` high); a surface raises it,\n  never lowers it: auth, payments, reward redemption, routing infrastructure, shared state → high.\n\n    ## Plan: \u003Cn>\n    Type: hotfix | change | new · Risk: low | medium | high — \u003Cwhy> · Surfaces: … · Files: …\n    Skills read: … · Design: \u003Cfigma url + tokens> | none\n    Steps: 1) … 2) …\n    Not changing: \u003Cwhat this plan deliberately leaves alone>\n    Evidence: \u003Cfile:line for each step's claim about the current code>\n    Tests: \u003Cwhich to update or add>\n    Risks \u002F open questions: … (each with the option you recommend)\n\n## 3. Work\n\n- **UI.** Build with the UI library this repo already uses, as its own UI skill describes; reuse its\n  components and theme before writing new ones. Never add the design system or another kit without a\n  ticket that says so.\n- **Tests.** Something earns a test when its breakage is invisible to typecheck and lint and\n  expensive in production (`docs\u002FTESTING.md` when the repo has one); a bug fix adds the regression test that fails before\n  the fix. Formatting is automatic at commit (lint-staged runs prettier); never hand-format.\n- **Effects and memos** list only the values they really read: never a dependency added to satisfy\n  a lint rule, never one rebuilt every render (an object literal, an unmemoised selector result).\n  `exhaustive-deps` is off on purpose; check a new memo once with\n  `pnpm exec eslint --rule '{\"react-hooks\u002Fexhaustive-deps\":\"error\"}' \u003Cfile>` and keep what matters.\n- **Comments** say why, never what: at most 2 lines; a directive is one line,\n  `\u002F\u002F eslint-disable-next-line \u003Crule> -- \u003Cwhy>`, never a block or `@ts-ignore`.\n- Update the docs `scripts\u002Fgit\u002Fdoc-map.txt` maps to the touched code, or give that commit a\n  `Docs-skip: \u003Cclass> — \u003Cwhy>` trailer when none is needed.\n\n## 4. Gate — one command, stop at the first failure\n\n    bash scripts\u002Fagent\u002Fgates.sh task \u003Clow|medium|high>\n\nLow: the code rules on the working tree (untracked files included), typecheck, lint. Medium: plus\nthe changed files' own tests (each changed source file at —% statements, branches,\nfunctions and lines when the repo measures coverage; `—` means it does not) and the build. High:\nplus the full suite — the only place it runs. `⚠` lines from scan-diff are re-render risks to fix or justify. Then one advisory line per\nroute the diff touches — route, what to look for, `(diff: file:line)` — for a person to try at\ndesktop 1440 · iPad landscape 1180 · iPad portrait 820 · mobile 390; `\u002Ffinalize` carries the list forward.\n\n## Standing traps\n\n- **Attribution trailers.** Agents add `Co-authored-by` by default; the commit-msg hook rejects an\n  AI one, so keep attribution off in settings instead of fighting the hook.\n- **A new package.** `deps-check.mjs` fails one with an install script, a name one letter from a popular package or\n  a licence outside `DEPS_LICENSES`. Pick another package first; a reviewed need is a dated row in\n  `docs\u002FDEPS-EXCEPTIONS.md`, or `pnpm.onlyBuiltDependencies` for an install script you have read.\n- **A key in the bundle.** Only public client keys belong in browser code: `build-scan.sh` fails a server secret\n  the bundler inlined. Move it behind the API; never allowlist it.\n- **A gate edited in the hook.** The hooks call `scripts\u002Fagent\u002Fgates.sh`; a gate changes in\n  `scripts\u002Fagent\u002Fagent.conf` and `docs\u002FGATES.md`, or the docs drift.\n\u003C!-- repo-traps:start -->\n\u003C!-- Repo-specific traps go here. `trs-app agents upgrade` keeps this block on a re-run. -->\n\u003C!-- repo-traps:end -->\n\n## Evidence, then stop\n\nEnd with: commands run and their exit codes, files changed (these are the task-owned paths `\u002Fflow`\nrecords), tests added, and what was **not** verified. Never commit, push, or skip hooks\n(`--no-verify`): the commit question belongs to `\u002Fflow`, and the developer answers it. Never\ninvoke `\u002Ffinalize` yourself. A worked plan is in `docs\u002FGATES.md`.\n\n## Related\n\nCLAUDE.md · docs\u002FGATES.md · `flow` (runs this) · `finalize` (the check) · `review` · `scan`\n","# Task — read, plan by type, work, gate\n\nRequest: $ARGUMENTS\n\nMost bad changes start with typing before reading the rule that governs the area. `\u002Ftask` is the\norder of work that prevents that: read first, plan as much as the work type needs, build, gate. A\nskill beats a doc; the source beats both, and the doc gets fixed. Every change follows the rules in\n`CLAUDE.md` and this repo's own skills. With `TASK_MODE=auto`, `\u002Fflow` runs this for every request\nand owns the ticket, the commit question and the ship; alone, `\u002Ftask` ends at a green gate.\n\n## 0. Type, then how much stops you\n\n| Type | Plan | Gates in step 4 |\n|---|---|---|\n| `hotfix` | none: print the read-only diagnosis (cause, `file:line`, the one fix) and go | `low` |\n| `change` | print the plan and continue; stop for approval only when a step touches auth, payments, routing infrastructure or shared state | `medium` |\n| `new` | approved before the first edit — Claude Code: call EnterPlanMode now, steps 1–3 read-only, ExitPlanMode, wait; Cursor, Codex, Antigravity: the tool's plan mode (`\u002Fplan`), or print the plan and wait for `go` | `high` |\n\nA hotfix adds no dependency or migration and stays inside one module (`no-deps,no-migrations,module-scope`); one that\nmust is a `change`. Confirm the type in one line. An edit before the stop a type requires is a failed run.\n\n## 1. Read what governs the change — before the plan\n\n| The change touches | Read first |\n|---|---|\n| _(no domain skills yet)_ | — |\n| a module's logic | that module's doc (see `scripts\u002Fgit\u002Fdoc-map.txt`), and its skill if it has one |\n| a component, hook or helper, or any UI | this repo's own UI skill (listed in CLAUDE.md § Skills) and the components it already has |\n| a gate, a hook, a commit or a branch name | `docs\u002FGATES.md`; `CLAUDE.md` § Rules |\n\u003C!-- repo-routes:start -->\n\u003C!-- repo-routes:end -->\n\nIf no row matches, say so and add a **proposed skill** to the plan (name, scope, first three\nrules); it lands as its own change, never in this task. A `figma.com` URL in the request, or a\nscreen change, with the Figma MCP connected: `figma:get_design_context` on the node (and\n`get_variable_defs` for tokens) goes under Evidence; not connected → `Design: Figma MCP not\nconnected` and continue.\n\n## 2. Classify and plan — read-only\n\n- **Surfaces** — a view (`src\u002Fmodules\u002F\u003CModule>\u002Fviews`), a module store (`store\u002F`), the router,\n  the platform slice (`src\u002F{lib,config,store}`), an adapter (`src\u002Fpackages`), copy (`—`), the build.\n- **Ships in** — web.\n- **Risk** — the type sets it (`hotfix` low, `change` medium, `new` high); a surface raises it,\n  never lowers it: auth, payments, reward redemption, routing infrastructure, shared state → high.\n\n    ## Plan: \u003Cn>\n    Type: hotfix | change | new · Risk: low | medium | high — \u003Cwhy> · Surfaces: … · Files: …\n    Skills read: … · Design: \u003Cfigma url + tokens> | none\n    Steps: 1) … 2) …\n    Not changing: \u003Cwhat this plan deliberately leaves alone>\n    Evidence: \u003Cfile:line for each step's claim about the current code>\n    Tests: \u003Cwhich to update or add>\n    Risks \u002F open questions: … (each with the option you recommend)\n\n## 3. Work\n\n- **UI.** Build with the UI library this repo already uses, as its own UI skill describes; reuse its\n  components and theme before writing new ones. Never add the design system or another kit without a\n  ticket that says so.\n- **Tests.** Something earns a test when its breakage is invisible to typecheck and lint and\n  expensive in production (`docs\u002FTESTING.md` when the repo has one); a bug fix adds the regression test that fails before\n  the fix. Formatting is automatic at commit (lint-staged runs prettier); never hand-format.\n- **Effects and memos** list only the values they really read: never a dependency added to satisfy\n  a lint rule, never one rebuilt every render (an object literal, an unmemoised selector result).\n  `exhaustive-deps` is off on purpose; check a new memo once with\n  `pnpm exec eslint --rule '{\"react-hooks\u002Fexhaustive-deps\":\"error\"}' \u003Cfile>` and keep what matters.\n- **Comments** say why, never what: at most 2 lines; a directive is one line,\n  `\u002F\u002F eslint-disable-next-line \u003Crule> -- \u003Cwhy>`, never a block or `@ts-ignore`.\n- Update the docs `scripts\u002Fgit\u002Fdoc-map.txt` maps to the touched code, or give that commit a\n  `Docs-skip: \u003Cclass> — \u003Cwhy>` trailer when none is needed.\n\n## 4. Gate — one command, stop at the first failure\n\n    bash scripts\u002Fagent\u002Fgates.sh task \u003Clow|medium|high>\n\nLow: the code rules on the working tree (untracked files included), typecheck, lint. Medium: plus\nthe changed files' own tests (each changed source file at —% statements, branches,\nfunctions and lines when the repo measures coverage; `—` means it does not) and the build. High:\nplus the full suite — the only place it runs. `⚠` lines from scan-diff are re-render risks to fix or justify. Then one advisory line per\nroute the diff touches — route, what to look for, `(diff: file:line)` — for a person to try at\ndesktop 1440 · iPad landscape 1180 · iPad portrait 820 · mobile 390; `\u002Ffinalize` carries the list forward.\n\n## Standing traps\n\n- **Attribution trailers.** Agents add `Co-authored-by` by default; the commit-msg hook rejects an\n  AI one, so keep attribution off in settings instead of fighting the hook.\n- **A new package.** `deps-check.mjs` fails one with an install script, a name one letter from a popular package or\n  a licence outside `DEPS_LICENSES`. Pick another package first; a reviewed need is a dated row in\n  `docs\u002FDEPS-EXCEPTIONS.md`, or `pnpm.onlyBuiltDependencies` for an install script you have read.\n- **A key in the bundle.** Only public client keys belong in browser code: `build-scan.sh` fails a server secret\n  the bundler inlined. Move it behind the API; never allowlist it.\n- **A gate edited in the hook.** The hooks call `scripts\u002Fagent\u002Fgates.sh`; a gate changes in\n  `scripts\u002Fagent\u002Fagent.conf` and `docs\u002FGATES.md`, or the docs drift.\n\u003C!-- repo-traps:start -->\n\u003C!-- Repo-specific traps go here. `trs-app agents upgrade` keeps this block on a re-run. -->\n\u003C!-- repo-traps:end -->\n\n## Evidence, then stop\n\nEnd with: commands run and their exit codes, files changed (these are the task-owned paths `\u002Fflow`\nrecords), tests added, and what was **not** verified. Never commit, push, or skip hooks\n(`--no-verify`): the commit question belongs to `\u002Fflow`, and the developer answers it. Never\ninvoke `\u002Ffinalize` yourself. A worked plan is in `docs\u002FGATES.md`.\n\n## Related\n\nCLAUDE.md · docs\u002FGATES.md · `flow` (runs this) · `finalize` (the check) · `review` · `scan`\n",{"name":5,"description":7,"disable-model-invocation":23},true,"master","a821ad5a10a86d16e095e16b866a121679b7b793",".claude\u002Fskills\u002Ftask","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova","https:\u002F\u002Fgithub.com\u002Ftherewardstore","2026-10-29 16:11:17",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":23},[32,35,38],{"tool":33,"command":34},"npx","npx skills add therewardstore\u002Fapplaudiq-embed-cordova --skill task",{"tool":36,"command":37},"bunx","bunx skills add therewardstore\u002Fapplaudiq-embed-cordova --skill task",{"tool":39,"command":40},"pnpm","pnpm dlx skills add therewardstore\u002Fapplaudiq-embed-cordova --skill task","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Ftask\u002FSKILL.md",[],[44,51,58,65,72,79],{"skillNo":45,"name":46,"nameZh":6,"description":47,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":48,"categories":49,"keywords":50,"publishedAt":18},"01M77N3VA856KABVHJ3ND6S1CG","scan","Orients an agent in this repo once per session: reads every skill and governing doc in full, checks what they claim against the filesystem, and reports what governs this repo and where CLAUDE.md has drifted. Use only when the user types \u002Fscan.","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Fscan\u002FSKILL.md",[],[],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":18},"01M77N3Q48ZBNV3NE8Q3H0JH86","review","Reviews a diff or a pull request line by line against this repo's rules and prints findings a reviewer would leave as comments, split into Must fix and Worth a look. Never edits, commits or approves. Use when the user types \u002Freview [\u003Crange> | \u003CPR number>].","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Freview\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":18},"01M77N3HJM6QKMYNDT7CB5N9X5","handoff","Writes the QA test plan for a finished change — what changed for the user, where to look, and the test cases — into the tracked change record, in a tester's words with no code, and comments it on the ticket. Use when \u002Fflow reaches it, or when the user types \u002Fhandoff after \u002Ffinalize said ready.","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Fhandoff\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":18},"01M77N3DG8M5CXRQ3XQ1FHSKRS","flow","Carries any request in this repo from the first message to a pull request: decides where the work belongs and which type it is, settles the ticket and branch, runs \u002Ftask, asks the developer before any commit, runs the second check, \u002Fhandoff and ship.sh, and ends with what the developer does next. Use for every request when TASK_MODE=auto, or when the user types \u002Fflow \u003Crequest>.","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Fflow\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M77N38CV5E3G7QDBTN7WCWCV","finalize","Judges a change from exit codes only: stage 1 on the proposed commit before the commit question, stage 2 on the committed sha, or the branch range. Never fixes, commits, pushes or asks. Runs when \u002Fflow reaches it, or when the user types \u002Ffinalize.","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.claude\u002Fskills\u002Ffinalize\u002FSKILL.md",[],[],{"skillNo":80,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":18},"01M77N37DRWS5W01J86252P157","https:\u002F\u002Fgithub.com\u002Ftherewardstore\u002Fapplaudiq-embed-cordova\u002Fblob\u002Fmaster\u002F.agents\u002Fskills\u002Ftask\u002FSKILL.md",[],[]]