[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7CTX1972KT0X0RXBTRCKHVB":3,"skill-same:01M7CTX1972KT0X0RXBTRCKHVB":64},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":45},"01M7CTX1972KT0X0RXBTRCKHVB","ponytail",null,"Use when implementation needs source receipts or a thorough reuse-before-build audit of existing code, installed packages and pinned official APIs. For a quick simpler-path recommendation use efficient.","amoai-tech\u002Fmdeai","amoai-tech","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F320616881?v=4","01M7CTJYGWP82J2HHMQC5NN9JP","amo","01M7CTJYHE5YQRGQSDJES3Z0QE",2,"MIT","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fponytail\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: ponytail\ndescription: >-\n  Use when implementation needs source receipts or a thorough reuse-before-build audit of existing code, installed packages and pinned official APIs. For a quick simpler-path recommendation use efficient.\n\n---\n\n# Ponytail — reuse before build\n\nOwn one question: **should this code exist?** `tasks` owns execution, `research` owns evidence\ngathering, and the stack skills own implementation. This skill owns the decision that comes first.\n\nUpstream: `references\u002Fofficial\u002Fponytail\u002Fponytail.md` (vendored read-only, pinned in `upstream.yaml`).\nThe ladder and discipline below are the MDE form of it; the source receipt is our addition.\n\n## Scope the complete change before minimizing\n\nRead the actual code and trace the affected user\u002Fruntime path before deciding on a solution.\nCheck every caller, test, fixture, configuration, export and persisted data flow the change can\naffect. Identify user-facing failures, unauthorized access and potential data loss. The smallest\nchange is the **smallest complete, safe change** — not a one-line patch that leaves other callers\nor regression tests broken. Preserve existing interfaces, validation and error handling when moving\nor consolidating code. End the report with what was not checked and any material remaining risk.\n\n## The ladder\n\nRead the task and the code it touches, trace the real flow end to end, **then** stop at the first\nrung that safely satisfies the requirement:\n\n1. Does this need to be built at all? (YAGNI)\n2. Does an existing repository helper, component, or pattern already do it?\n3. Does the standard library already do this?\n4. Does a native platform capability cover it — including an official dashboard or CLI surface?\n5. Does an already-installed dependency solve it?\n6. Does an official template, example, recipe, or CLI command cover it?\n7. Only then: write the minimum code that works.\n\nThe ladder runs *after* understanding, never instead of it. Rungs 3–6 are where most tasks are won:\nthe platform and the dependency list usually already ship the answer.\n\nChoose the earliest rung that satisfies **correctness, security, testability, reproducibility,\nversion compatibility, and automation**. A dashboard-only configuration is not sufficient when it\nmust be reproducible from Git — a setting that has to replay in every environment belongs in a\nmigration or committed config, not a console.\n\nDo not build a custom abstraction merely because it is easy to write. Do not add speculative\noptions, wrappers, conversions, configuration, dependencies or \"for later\" code. A one-liner is\ngood only when the next reader can understand it at a glance; prefer the clearer small solution.\n\n**Not lazy about:** understanding the problem first, input validation at trust boundaries, error\nhandling that prevents data loss, security, accessibility, and anything explicitly requested. Never\ntrade any of these for a smaller diff.\n\n## Efficient implementation order\n\nFor substantial work, before writing custom code:\n\n1. Check whether the requirement needs to exist.\n2. Search this repository for an existing implementation.\n3. Check installed packages and platform-native capabilities.\n4. Check official dashboard, CLI, and configuration surfaces where relevant.\n5. Read the official documentation for the exact installed version.\n6. Inspect the official repository or a working example when behaviour matters.\n7. Prefer an official template, example, recipe, CLI command, or prebuilt module.\n8. Adapt the smallest supported solution.\n9. Write custom code only for the remaining gap.\n10. Leave a runnable verification proving the chosen solution.\n\nSteps 5–7 are also where the source receipt comes from: the source you actually used is the one\nworth recording.\n\n## Bug fixes climb from the root\n\nA report names a symptom. Grep every caller of the function you are about to touch and fix the\nshared function once. Patching only the path named in the ticket leaves sibling callers broken, and\none guard in the shared function is a smaller diff than one guard per caller.\n\n## Deliberate ceilings get a comment\n\nWhen you knowingly cut a corner — a global lock, an O(n²) scan, a naive heuristic — mark it:\n\n```ts\n\u002F\u002F ponytail: global lock; upgrade to per-key locks if contention shows up in metrics.\n```\n\nName the ceiling and the upgrade path. An unmarked shortcut is indistinguishable from an oversight.\nUpstream Ponytail 5 calls this marker `shortcut:`; MDE deliberately retains its existing\n`ponytail:` convention so current code and tools keep a single searchable marker.\n\n## One runnable check\n\nNon-trivial logic leaves **one** runnable check behind: the smallest thing that fails if the logic\nbreaks. An assert-based self-check or one small test file is enough — no frameworks, no fixtures.\nTrivial one-liners need none. Delete the check only when you delete the logic.\n\n## Source receipts\n\nReading a doc is not reusing it. An external source that **materially affects the implementation or\nthe decision** produces one row — in the PR body, or in the task's evidence file. Background reading\ngets none: an inflated citation table is documentation noise, and it buries the rows that mattered.\nA source with no row did not influence the work, so it must not be cited as its justification.\n\nThe required fields are the contract in `AGENTS.md` § Evidence rules → Source receipts. This skill\nowns the judgement, which is the part that gets misused:\n\n- **COPY** — verbatim. Keeps its license\u002Fattribution and takes no local edits.\n- **ADAPT** — same approach, changed to our conventions. State what changed.\n- **MODEL** — follow the shape, write our own. State what we deliberately diverged on.\n- **REFERENCE ONLY** — ships nothing, therefore it **requires a stated reason** plus the alternative\n  you rejected. An unexplained `REFERENCE ONLY` is how this rule becomes decorative: it lets any\n  amount of reading stand in for a decision.\n\nA receipt is cheap. The failure it prevents is expensive: six weeks later nobody can tell whether a\npattern was copied from a current official example or recalled from memory.\n\n## Two kinds of proof\n\nDo not confuse verifying the source with verifying our use of it.\n\n- **Source proof** — the API, behaviour, version, security property, or platform capability is\n  confirmed against a primary source.\n- **Implementation proof** — this repository is confirmed to use it correctly.\n\nOne never substitutes for the other. A URL is not a test; a passing test does not validate an\nassumption nobody checked. Both are required, and the difference is where this repository's most\nexpensive mistakes have come from.\n\n## Verify the current contract before copying an example\n\nDetermine what this repository actually uses, in this order:\n\n1. the installed package or pinned version (`node_modules\u002F`, lockfile);\n2. current source and types in this repo;\n3. official documentation **for that version**;\n4. official repository examples;\n5. third-party examples, only when primary sources are insufficient.\n\nOur pins lag upstream more often than not. A current upstream example applied to an older pinned API\nis a new bug wearing a citation.\n\n\n## Every task opens with one written line\n\nBefore the first edit, answer in one line: *is there a faster, smaller, or better-supported way to do\nthis than building it?* Name the rung you stopped at. When the answer is \"write it ourselves\", name\nthe rungs you checked and why each failed. This is the whole rule in practice — the ladder is only\nreal if the rung is stated.\n\n## Example — adding a UI component\n\n| Bad sequence | Preferred sequence |\n| -- | -- |\n| requirements → write a custom modal → debug focus trapping → debug keyboard navigation → patch accessibility → ship | requirements → search existing MDE components → check the installed UI package → check the official component recipe → adapt the accessible component → add only MDE-specific behaviour → Playwright verification |\n\nThe preferred sequence is shorter *and* lands more accessible, because focus handling and keyboard\nnavigation were already solved by the component that shipped.\n\n## Worked example (this repo)\n\nTask: stop an unreviewed migration reaching production before `supabase db push`.\n\n- Rung 1 — does a check need to exist? Yes: a live near-miss had already happened.\n- Rung 2 — does the repo already have it? No guard existed; what was reusable was the CLI's own rule.\n- Rung 3\u002F4 — **the platform already implements the rule.** `supabase db push` decides which files it\n  applies via `ListLocalMigrations`, so the guard mirrors that instead of inventing a heuristic.\n\nThe first version asked Git \"which files are untracked?\" — rung 3 skipped, and it was wrong in both\ndirections (a gitignored migration was applied while the guard said PASS; an uppercase `.SQL` was\nreported but never applied). Mirroring the CLI's own rule fixed both.\n\nReceipt — the Destination is an **open pull request, not `main`**, so that path does not resolve in\nthis checkout yet. A receipt may name unlanded work, but it must say so:\n\n| Field | Value |\n| -- | -- |\n| URL | `https:\u002F\u002Fgithub.com\u002Fsupabase\u002Fcli\u002Fblob\u002Fdevelop\u002Fapps\u002Fcli-go\u002Fpkg\u002Fmigration\u002Flist.go` |\n| Source | `ListLocalMigrations` (`fs.ReadDir`, top level only, directories skipped) and `migrateFilePattern` in `apps\u002Fcli-go\u002Fpkg\u002Fmigration\u002Ffile.go` |\n| Decision | Which files does `supabase db push` actually apply, and how must the guard select the same set? |\n| Disposition | ADAPT |\n| Destination | `scripts\u002Fpreflight-migration-release.mjs` — **in PR #147, not yet on `main`** |\n| Implementation | Mirror the CLI filename rule as `CLI_MIGRATION_FILE_PATTERN` + `listPushableMigrations` |\n| Verification | `node --test scripts\u002F__tests__\u002Fpreflight-migration-release.test.mjs`, plus a live `db push --dry-run` on the planted-file fixture |\n| Version \u002F commit | Supabase CLI 2.118.0 as installed (this repo does not pin it); rule read at `develop` |\n\n## References\n\n- `references\u002Fsource-map.md` — the authoritative source to check per area, and how to use it.\n- `references\u002Fofficial\u002Fponytail\u002Fponytail.md` — vendored upstream rule (read-only).\n\n## Verification\n\nThe rule is satisfied when the diff is the smallest that safely works **and** every consulted source\nhas a receipt with a real destination and a real verification command. A task with no receipts and no\nstated rung has not shown that it climbed the ladder — it has only not mentioned it.\n","# Ponytail — reuse before build\n\nOwn one question: **should this code exist?** `tasks` owns execution, `research` owns evidence\ngathering, and the stack skills own implementation. This skill owns the decision that comes first.\n\nUpstream: `references\u002Fofficial\u002Fponytail\u002Fponytail.md` (vendored read-only, pinned in `upstream.yaml`).\nThe ladder and discipline below are the MDE form of it; the source receipt is our addition.\n\n## Scope the complete change before minimizing\n\nRead the actual code and trace the affected user\u002Fruntime path before deciding on a solution.\nCheck every caller, test, fixture, configuration, export and persisted data flow the change can\naffect. Identify user-facing failures, unauthorized access and potential data loss. The smallest\nchange is the **smallest complete, safe change** — not a one-line patch that leaves other callers\nor regression tests broken. Preserve existing interfaces, validation and error handling when moving\nor consolidating code. End the report with what was not checked and any material remaining risk.\n\n## The ladder\n\nRead the task and the code it touches, trace the real flow end to end, **then** stop at the first\nrung that safely satisfies the requirement:\n\n1. Does this need to be built at all? (YAGNI)\n2. Does an existing repository helper, component, or pattern already do it?\n3. Does the standard library already do this?\n4. Does a native platform capability cover it — including an official dashboard or CLI surface?\n5. Does an already-installed dependency solve it?\n6. Does an official template, example, recipe, or CLI command cover it?\n7. Only then: write the minimum code that works.\n\nThe ladder runs *after* understanding, never instead of it. Rungs 3–6 are where most tasks are won:\nthe platform and the dependency list usually already ship the answer.\n\nChoose the earliest rung that satisfies **correctness, security, testability, reproducibility,\nversion compatibility, and automation**. A dashboard-only configuration is not sufficient when it\nmust be reproducible from Git — a setting that has to replay in every environment belongs in a\nmigration or committed config, not a console.\n\nDo not build a custom abstraction merely because it is easy to write. Do not add speculative\noptions, wrappers, conversions, configuration, dependencies or \"for later\" code. A one-liner is\ngood only when the next reader can understand it at a glance; prefer the clearer small solution.\n\n**Not lazy about:** understanding the problem first, input validation at trust boundaries, error\nhandling that prevents data loss, security, accessibility, and anything explicitly requested. Never\ntrade any of these for a smaller diff.\n\n## Efficient implementation order\n\nFor substantial work, before writing custom code:\n\n1. Check whether the requirement needs to exist.\n2. Search this repository for an existing implementation.\n3. Check installed packages and platform-native capabilities.\n4. Check official dashboard, CLI, and configuration surfaces where relevant.\n5. Read the official documentation for the exact installed version.\n6. Inspect the official repository or a working example when behaviour matters.\n7. Prefer an official template, example, recipe, CLI command, or prebuilt module.\n8. Adapt the smallest supported solution.\n9. Write custom code only for the remaining gap.\n10. Leave a runnable verification proving the chosen solution.\n\nSteps 5–7 are also where the source receipt comes from: the source you actually used is the one\nworth recording.\n\n## Bug fixes climb from the root\n\nA report names a symptom. Grep every caller of the function you are about to touch and fix the\nshared function once. Patching only the path named in the ticket leaves sibling callers broken, and\none guard in the shared function is a smaller diff than one guard per caller.\n\n## Deliberate ceilings get a comment\n\nWhen you knowingly cut a corner — a global lock, an O(n²) scan, a naive heuristic — mark it:\n\n```ts\n\u002F\u002F ponytail: global lock; upgrade to per-key locks if contention shows up in metrics.\n```\n\nName the ceiling and the upgrade path. An unmarked shortcut is indistinguishable from an oversight.\nUpstream Ponytail 5 calls this marker `shortcut:`; MDE deliberately retains its existing\n`ponytail:` convention so current code and tools keep a single searchable marker.\n\n## One runnable check\n\nNon-trivial logic leaves **one** runnable check behind: the smallest thing that fails if the logic\nbreaks. An assert-based self-check or one small test file is enough — no frameworks, no fixtures.\nTrivial one-liners need none. Delete the check only when you delete the logic.\n\n## Source receipts\n\nReading a doc is not reusing it. An external source that **materially affects the implementation or\nthe decision** produces one row — in the PR body, or in the task's evidence file. Background reading\ngets none: an inflated citation table is documentation noise, and it buries the rows that mattered.\nA source with no row did not influence the work, so it must not be cited as its justification.\n\nThe required fields are the contract in `AGENTS.md` § Evidence rules → Source receipts. This skill\nowns the judgement, which is the part that gets misused:\n\n- **COPY** — verbatim. Keeps its license\u002Fattribution and takes no local edits.\n- **ADAPT** — same approach, changed to our conventions. State what changed.\n- **MODEL** — follow the shape, write our own. State what we deliberately diverged on.\n- **REFERENCE ONLY** — ships nothing, therefore it **requires a stated reason** plus the alternative\n  you rejected. An unexplained `REFERENCE ONLY` is how this rule becomes decorative: it lets any\n  amount of reading stand in for a decision.\n\nA receipt is cheap. The failure it prevents is expensive: six weeks later nobody can tell whether a\npattern was copied from a current official example or recalled from memory.\n\n## Two kinds of proof\n\nDo not confuse verifying the source with verifying our use of it.\n\n- **Source proof** — the API, behaviour, version, security property, or platform capability is\n  confirmed against a primary source.\n- **Implementation proof** — this repository is confirmed to use it correctly.\n\nOne never substitutes for the other. A URL is not a test; a passing test does not validate an\nassumption nobody checked. Both are required, and the difference is where this repository's most\nexpensive mistakes have come from.\n\n## Verify the current contract before copying an example\n\nDetermine what this repository actually uses, in this order:\n\n1. the installed package or pinned version (`node_modules\u002F`, lockfile);\n2. current source and types in this repo;\n3. official documentation **for that version**;\n4. official repository examples;\n5. third-party examples, only when primary sources are insufficient.\n\nOur pins lag upstream more often than not. A current upstream example applied to an older pinned API\nis a new bug wearing a citation.\n\n\n## Every task opens with one written line\n\nBefore the first edit, answer in one line: *is there a faster, smaller, or better-supported way to do\nthis than building it?* Name the rung you stopped at. When the answer is \"write it ourselves\", name\nthe rungs you checked and why each failed. This is the whole rule in practice — the ladder is only\nreal if the rung is stated.\n\n## Example — adding a UI component\n\n| Bad sequence | Preferred sequence |\n| -- | -- |\n| requirements → write a custom modal → debug focus trapping → debug keyboard navigation → patch accessibility → ship | requirements → search existing MDE components → check the installed UI package → check the official component recipe → adapt the accessible component → add only MDE-specific behaviour → Playwright verification |\n\nThe preferred sequence is shorter *and* lands more accessible, because focus handling and keyboard\nnavigation were already solved by the component that shipped.\n\n## Worked example (this repo)\n\nTask: stop an unreviewed migration reaching production before `supabase db push`.\n\n- Rung 1 — does a check need to exist? Yes: a live near-miss had already happened.\n- Rung 2 — does the repo already have it? No guard existed; what was reusable was the CLI's own rule.\n- Rung 3\u002F4 — **the platform already implements the rule.** `supabase db push` decides which files it\n  applies via `ListLocalMigrations`, so the guard mirrors that instead of inventing a heuristic.\n\nThe first version asked Git \"which files are untracked?\" — rung 3 skipped, and it was wrong in both\ndirections (a gitignored migration was applied while the guard said PASS; an uppercase `.SQL` was\nreported but never applied). Mirroring the CLI's own rule fixed both.\n\nReceipt — the Destination is an **open pull request, not `main`**, so that path does not resolve in\nthis checkout yet. A receipt may name unlanded work, but it must say so:\n\n| Field | Value |\n| -- | -- |\n| URL | `https:\u002F\u002Fgithub.com\u002Fsupabase\u002Fcli\u002Fblob\u002Fdevelop\u002Fapps\u002Fcli-go\u002Fpkg\u002Fmigration\u002Flist.go` |\n| Source | `ListLocalMigrations` (`fs.ReadDir`, top level only, directories skipped) and `migrateFilePattern` in `apps\u002Fcli-go\u002Fpkg\u002Fmigration\u002Ffile.go` |\n| Decision | Which files does `supabase db push` actually apply, and how must the guard select the same set? |\n| Disposition | ADAPT |\n| Destination | `scripts\u002Fpreflight-migration-release.mjs` — **in PR #147, not yet on `main`** |\n| Implementation | Mirror the CLI filename rule as `CLI_MIGRATION_FILE_PATTERN` + `listPushableMigrations` |\n| Verification | `node --test scripts\u002F__tests__\u002Fpreflight-migration-release.test.mjs`, plus a live `db push --dry-run` on the planted-file fixture |\n| Version \u002F commit | Supabase CLI 2.118.0 as installed (this repo does not pin it); rule read at `develop` |\n\n## References\n\n- `references\u002Fsource-map.md` — the authoritative source to check per area, and how to use it.\n- `references\u002Fofficial\u002Fponytail\u002Fponytail.md` — vendored upstream rule (read-only).\n\n## Verification\n\nThe rule is satisfied when the diff is the smallest that safely works **and** every consulted source\nhas a receipt with a real destination and a real verification command. A task with no receipts and no\nstated rung has not shown that it climbed the ladder — it has only not mentioned it.\n",{"name":5,"description":7},"main","9955f6a8aa5de3c14fea562c7e6b72973366ba3c",".claude\u002Fskills\u002Fponytail",1,"https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai","https:\u002F\u002Fmdeai-psi.vercel.app","https:\u002F\u002Fgithub.com\u002Famoai-tech","2026-11-13 03:54:16",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add amoai-tech\u002Fmdeai --skill ponytail",{"tool":38,"command":39},"bunx","bunx skills add amoai-tech\u002Fmdeai --skill ponytail",{"tool":41,"command":42},"pnpm","pnpm dlx skills add amoai-tech\u002Fmdeai --skill ponytail","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fponytail\u002FSKILL.md",true,[46,52,56,60],{"path":47,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":50,"sourceHtmlUrl":51},"evals\u002Fevals.json","REFERENCE","参考文档",3674,"https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fponytail\u002Fevals\u002Fevals.json",{"path":53,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":54,"sourceHtmlUrl":55},"references\u002Fofficial\u002Fponytail\u002Fponytail.md",2623,"https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fponytail\u002Freferences\u002Fofficial\u002Fponytail\u002Fponytail.md",{"path":57,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":58,"sourceHtmlUrl":59},"references\u002Fsource-map.md",8952,"https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fponytail\u002Freferences\u002Fsource-map.md",{"path":61,"resourceType":48,"resourceTypeDesc":49,"sizeBytes":62,"sourceHtmlUrl":63},"upstream.yaml",857,"https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fponytail\u002Fupstream.yaml",[65,72,79,86,93,100],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7CV3D59CQX8N178NN3W5DEQ","mde-pr-review-supabase","Use when reviewing MDE Supabase SQL, Auth and RLS changes in a pull request.","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002Fscripts\u002Fpr-agent\u002Fskills\u002Fsupabase\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7CV39XR1SPH7C4FNXAXMM36","mde-pr-review-stripe","Use when reviewing MDE Stripe payments and webhooks changes in a pull request.","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002Fscripts\u002Fpr-agent\u002Fskills\u002Fstripe\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7CV37HHNM5A144SEX2VE05N","mde-pr-review-nextjs","Use when reviewing MDE Next.js App Router changes in a pull request.","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002Fscripts\u002Fpr-agent\u002Fskills\u002Fnextjs\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M7CV34C1M5STTKHX5CGVN3VS","mde-pr-review-mastra","Use when reviewing MDE Mastra agent, tool and workflow changes in a pull request.","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002Fscripts\u002Fpr-agent\u002Fskills\u002Fmastra\u002FSKILL.md",[],[],{"skillNo":94,"name":95,"nameZh":6,"description":96,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":97,"categories":98,"keywords":99,"publishedAt":19},"01M7CV31A37VPRRASAVGE8K8SN","mde-pr-review-maps","Use when reviewing MDE Google Maps and Places changes in a pull request.","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002Fscripts\u002Fpr-agent\u002Fskills\u002Fmaps\u002FSKILL.md",[],[],{"skillNo":101,"name":102,"nameZh":6,"description":103,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":104,"categories":105,"keywords":106,"publishedAt":19},"01M7CV2Y70DQJ1VXFHS732CSQ4","mde-pr-review-copilotkit","Use when reviewing MDE CopilotKit and AG-UI changes in a pull request.","https:\u002F\u002Fgithub.com\u002Famoai-tech\u002Fmdeai\u002Fblob\u002Fmain\u002Fscripts\u002Fpr-agent\u002Fskills\u002Fcopilotkit\u002FSKILL.md",[],[]]