[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75ME522XS22DQACWMBM5PS2":3,"skill-same:01M75ME522XS22DQACWMBM5PS2":45},{"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":23,"commitSha":24,"skillDirectory":25,"forksCount":26,"repositoryHtmlUrl":27,"repositoryHomepageUrl":28,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M75ME522XS22DQACWMBM5PS2","adversarial-review",null,"Use before claiming a high-stakes artifact done, verified, or merged. Have an independent grader adversarially review it — prompted to refute and find reasons not to ship, not to bless — and iterate to an explicit ship-or-revise verdict.","selamy-labs\u002Fagent-skills","selamy-labs","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F287832944?v=4","01M75MDKJ0CKAM94KZ7TWQJM5Q","01M75MDKJNP3TATHAW1MX1FWM6",0,"MIT","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fadversarial-review\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: adversarial-review\ndescription: Use before claiming a high-stakes artifact done, verified, or merged. Have an independent grader adversarially review it — prompted to refute and find reasons not to ship, not to bless — and iterate to an explicit ship-or-revise verdict.\n---\n\n# Adversarial Review\n\nThe author is the worst judge of their own work — they review it with the same\nassumptions that produced it. Before you call a high-stakes artifact done,\nverified, or merged, hand it to an **independent grader whose job is to refute\nit**: find the reason it should not ship. What survives a genuine attempt to\nbreak it is far more trustworthy than what you blessed yourself.\n\n## Refute, don't bless\n\nThe grader's task is adversarial: *\"Here is a diff\u002Fartifact claimed ready. Find\nthe strongest reasons NOT to ship it — bugs, missed cases, unsafe assumptions,\nunmet requirements. Default to REVISE if uncertain.\"* A prompt that asks \"does\nthis look good?\" gets a rubber stamp; a prompt that asks \"why is this wrong?\"\ngets the defects. Uncertainty is a REVISE, not a SHIP — a weak \"seems fine\"\nguards nothing.\n\n## Independence is the whole point\n\nA grader that shares the author's context inherits the author's blind spots.\nMaximize independence:\n\n- **Different context\u002Fmodel** than the one that produced the artifact — not the\n  author re-reading their own work.\n- **Clean-room where possible**: review the diff against an *untouched* checkout\n  of the base, so the grader judges what actually changed, not the author's\n  narrative about it.\n- **Anti-reward-hack**: the grader is graded on finding real defects, not on\n  agreeing. It does not get to clear the artifact by restating its intent.\n\n## Explicit verdict, then iterate\n\nThe output is a decision, not a vibe: **SHIP** or **REVISE**, with the specific\ndefects that drove it. REVISE → fix the named defects → re-review. Loop until a\nclean SHIP. Record the verdict with the artifact so \"it was reviewed\" is a fact,\nnot a memory.\n\n## Scale to the stakes\n\nAdversarial review costs an extra cycle; spend it where being wrong is expensive:\nlive-money or financial changes, security\u002Fauth\u002Fpermission edits, fleet-wide or\nirreversible infrastructure, data migrations, anything you cannot easily roll\nback. A typo fix does not need a tribunal. Match the rigor to the blast radius.\n\n## Make it standing, not ad-hoc\n\nA review you *remember* to run is a review you will skip under pressure. Wire it\nto the boundary where the claim is made — before a done-mark, before opening the\nPR — with a hook or checklist gate, so the adversarial review is part of the\ndefinition of done rather than an optional courtesy. For higher assurance, use\nseveral independent graders with distinct lenses (correctness, security, does it\nactually reproduce) and require a majority to clear.\n\n## Where it fits\n\nAdversarial review answers *\"is this actually right and safe to ship?\"* It pairs\nwith two siblings: **verify the real artifact** (does the thing exist and behave\nin the closest-to-production form?) and **process-aware done** (never mark done\non a proxy). Run the adversarial review, verify the real artifact it points at,\nthen — and only then — claim done.\n","# Adversarial Review\n\nThe author is the worst judge of their own work — they review it with the same\nassumptions that produced it. Before you call a high-stakes artifact done,\nverified, or merged, hand it to an **independent grader whose job is to refute\nit**: find the reason it should not ship. What survives a genuine attempt to\nbreak it is far more trustworthy than what you blessed yourself.\n\n## Refute, don't bless\n\nThe grader's task is adversarial: *\"Here is a diff\u002Fartifact claimed ready. Find\nthe strongest reasons NOT to ship it — bugs, missed cases, unsafe assumptions,\nunmet requirements. Default to REVISE if uncertain.\"* A prompt that asks \"does\nthis look good?\" gets a rubber stamp; a prompt that asks \"why is this wrong?\"\ngets the defects. Uncertainty is a REVISE, not a SHIP — a weak \"seems fine\"\nguards nothing.\n\n## Independence is the whole point\n\nA grader that shares the author's context inherits the author's blind spots.\nMaximize independence:\n\n- **Different context\u002Fmodel** than the one that produced the artifact — not the\n  author re-reading their own work.\n- **Clean-room where possible**: review the diff against an *untouched* checkout\n  of the base, so the grader judges what actually changed, not the author's\n  narrative about it.\n- **Anti-reward-hack**: the grader is graded on finding real defects, not on\n  agreeing. It does not get to clear the artifact by restating its intent.\n\n## Explicit verdict, then iterate\n\nThe output is a decision, not a vibe: **SHIP** or **REVISE**, with the specific\ndefects that drove it. REVISE → fix the named defects → re-review. Loop until a\nclean SHIP. Record the verdict with the artifact so \"it was reviewed\" is a fact,\nnot a memory.\n\n## Scale to the stakes\n\nAdversarial review costs an extra cycle; spend it where being wrong is expensive:\nlive-money or financial changes, security\u002Fauth\u002Fpermission edits, fleet-wide or\nirreversible infrastructure, data migrations, anything you cannot easily roll\nback. A typo fix does not need a tribunal. Match the rigor to the blast radius.\n\n## Make it standing, not ad-hoc\n\nA review you *remember* to run is a review you will skip under pressure. Wire it\nto the boundary where the claim is made — before a done-mark, before opening the\nPR — with a hook or checklist gate, so the adversarial review is part of the\ndefinition of done rather than an optional courtesy. For higher assurance, use\nseveral independent graders with distinct lenses (correctness, security, does it\nactually reproduce) and require a majority to clear.\n\n## Where it fits\n\nAdversarial review answers *\"is this actually right and safe to ship?\"* It pairs\nwith two siblings: **verify the real artifact** (does the thing exist and behave\nin the closest-to-production form?) and **process-aware done** (never mark done\non a proxy). Run the adversarial review, verify the real artifact it points at,\nthen — and only then — claim done.\n",{"name":5,"description":7},"main","54007b08157feb09e5bb9d7ff5ce4ddd21625dbf","skills\u002Fadversarial-review",3,"https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills","https:\u002F\u002Fselamy.dev","https:\u002F\u002Fgithub.com\u002Fselamy-labs","2026-11-15 03:50:58",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add selamy-labs\u002Fagent-skills --skill adversarial-review",{"tool":37,"command":38},"bunx","bunx skills add selamy-labs\u002Fagent-skills --skill adversarial-review",{"tool":40,"command":41},"pnpm","pnpm dlx skills add selamy-labs\u002Fagent-skills --skill adversarial-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Fskills\u002Fadversarial-review\u002FSKILL.md",true,[],[46,53,60,67,73,80],{"skillNo":47,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":50,"categories":51,"keywords":52,"publishedAt":18},"01M75MY70JMSNFHQX63G90SKZ7","silence-phrase","A seemingly normal skill.","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fsilence-phrase\u002FSKILL.md",[],[],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":18},"01M75MY5PKVKY60M8H2DFTW58C","selamy-bad-skill","This skill has a vendor prefix that should be rejected.","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fprefixed-name\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":18},"01M75MY1TKXCH6WYN9ET0W4T02","oversized-description","This is a very long description that contains way too many words and should be rejected by the validator because it exceeds the forty five word limit that is enforced by the naming lint checker which ensures skill descriptions stay concise and readable for humans and machines alike without unnecessary verbosity or padding","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Foversized-description\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"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":70,"categories":71,"keywords":72,"publishedAt":18},"01M75MY0SZ7EKT99NWJEJ8HKS2","missing-frontmatter","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fmissing-frontmatter\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":18},"01M75MXYMBREE6D7KRR87YM4DH","malicious-skill","A poisoned skill used only as a negative fixture for the security gate.","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fmalicious-skill\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":18},"01M75MXX2B7AC2QSSRRQQXTA0N","leaking-org-name","A skill about something.","https:\u002F\u002Fgithub.com\u002Fselamy-labs\u002Fagent-skills\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fleaking-org-name\u002FSKILL.md",[],[]]