[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7ANVBFAMW1WKD2YV3JYNJWW":3,"skill-same:01M7ANVBFAMW1WKD2YV3JYNJWW":46},{"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":6,"ownerHtmlUrl":28,"ownerBlogUrl":29,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M7ANVBFAMW1WKD2YV3JYNJWW","postmortem",null,"Writing up an incident after it is resolved — assembling the timeline from logs and messages, separating what happened from why, and producing follow-up actions that name an owner and a trigger rather than an intention.","juchengquan\u002Fkingfisher","juchengquan","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F34294876?v=4","01M7ANV2BHYZVSJSXVC0C1PF6B","01M7ANV2C4G57YS16Y5NCTEVNN",0,"MIT","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fskills\u002Fincident\u002Fpostmortem\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: postmortem\ndescription: Writing up an incident after it is resolved — assembling the timeline from logs and messages, separating what happened from why, and producing follow-up actions that name an owner and a trigger rather than an intention.\n---\n\n# Postmortem\n\n## When to use\n\nA task asks you to write up an incident, outage or regression that has already\nbeen resolved, or to turn a pile of logs and chat messages into an account\nsomebody can act on.\n\nNot for triaging something still happening. That wants the shortest path to a\nmitigation, and this asks you to slow down and reconstruct.\n\n## Where this one lives\n\nThis skill sits in `skills\u002Fincident\u002F` rather than directly under `skills\u002F`, which\nis the only structural thing it is here to show. A folder under the skills root\nis a **source**, and `incident` is one — so this skill can be written\n`incident::postmortem` as well as `postmortem`.\n\nBoth spellings resolve, and the short one is what the agent is offered while the\nname is unique. The qualifier exists for the day it is not: two parties may each\nship a `postmortem`, and then naming it bare is refused rather than guessed —\nwhich turns a working grant into a loud error instead of quietly changing which\nprocedure the caller gets.\n\nOne folder of grouping and no more. Tools and subagents nest as deep as you like;\nskills do not, because the agent reads them off the filesystem itself and looks\nexactly one level down. `skills\u002Fincident\u002Fdeep\u002Fpostmortem\u002F` would be invisible,\nand `kingfisher list` reports anything hiding below that rather than leaving the\ncatalogue looking empty.\n\n## The procedure\n\n**Build the timeline first, and only then explain it.** Work from timestamps you\ncan point at — deploy records, log lines, message times. Write each entry as\n*time, what was observed, who observed it*. Resist writing causes into this\nsection; an incident where the first theory was wrong is the ordinary case, and a\ntimeline that already assumes the answer hides the minutes where nobody knew it.\n\n**Separate the trigger from the cause.** The trigger is what happened on the day;\nthe cause is the condition that made the trigger sufficient. A deploy that\nexposed a latent limit is one trigger and one cause, and a writeup that names\nonly the deploy leaves the limit in place for the next one.\n\n**Say what detection actually was.** Whether the alert fired, or whether a person\nnoticed. Time-to-detect is the number most writeups round in their own favour,\nand it is the one that predicts the next incident's length.\n\n**Write actions with an owner and a trigger, not an intention.** \"Add a limit\nalert, owned by the platform team, before the next capacity change\" is an action.\n\"Improve monitoring\" is a sentiment. An action nobody is named on is one nobody\ndoes.\n\n**Name what went right.** A rollback that worked, a runbook that was accurate, a\ndashboard that showed the problem. These are the parts worth protecting when\nsomebody later proposes removing them.\n\n## What to hand back\n\nA timeline, a trigger, a cause, detection and recovery times, and the actions.\nSay plainly where the record is thin — an unlogged window is a finding about the\nlogging, and an account that quietly smooths over it teaches nobody anything.\n","# Postmortem\n\n## When to use\n\nA task asks you to write up an incident, outage or regression that has already\nbeen resolved, or to turn a pile of logs and chat messages into an account\nsomebody can act on.\n\nNot for triaging something still happening. That wants the shortest path to a\nmitigation, and this asks you to slow down and reconstruct.\n\n## Where this one lives\n\nThis skill sits in `skills\u002Fincident\u002F` rather than directly under `skills\u002F`, which\nis the only structural thing it is here to show. A folder under the skills root\nis a **source**, and `incident` is one — so this skill can be written\n`incident::postmortem` as well as `postmortem`.\n\nBoth spellings resolve, and the short one is what the agent is offered while the\nname is unique. The qualifier exists for the day it is not: two parties may each\nship a `postmortem`, and then naming it bare is refused rather than guessed —\nwhich turns a working grant into a loud error instead of quietly changing which\nprocedure the caller gets.\n\nOne folder of grouping and no more. Tools and subagents nest as deep as you like;\nskills do not, because the agent reads them off the filesystem itself and looks\nexactly one level down. `skills\u002Fincident\u002Fdeep\u002Fpostmortem\u002F` would be invisible,\nand `kingfisher list` reports anything hiding below that rather than leaving the\ncatalogue looking empty.\n\n## The procedure\n\n**Build the timeline first, and only then explain it.** Work from timestamps you\ncan point at — deploy records, log lines, message times. Write each entry as\n*time, what was observed, who observed it*. Resist writing causes into this\nsection; an incident where the first theory was wrong is the ordinary case, and a\ntimeline that already assumes the answer hides the minutes where nobody knew it.\n\n**Separate the trigger from the cause.** The trigger is what happened on the day;\nthe cause is the condition that made the trigger sufficient. A deploy that\nexposed a latent limit is one trigger and one cause, and a writeup that names\nonly the deploy leaves the limit in place for the next one.\n\n**Say what detection actually was.** Whether the alert fired, or whether a person\nnoticed. Time-to-detect is the number most writeups round in their own favour,\nand it is the one that predicts the next incident's length.\n\n**Write actions with an owner and a trigger, not an intention.** \"Add a limit\nalert, owned by the platform team, before the next capacity change\" is an action.\n\"Improve monitoring\" is a sentiment. An action nobody is named on is one nobody\ndoes.\n\n**Name what went right.** A rollback that worked, a runbook that was accurate, a\ndashboard that showed the problem. These are the parts worth protecting when\nsomebody later proposes removing them.\n\n## What to hand back\n\nA timeline, a trigger, a cause, detection and recovery times, and the actions.\nSay plainly where the record is thin — an unlogged window is a finding about the\nlogging, and an account that quietly smooths over it teaches nobody anything.\n",{"name":5,"description":7},"main","93f4568cbc546f8e77663eeeef321f13c612a73c","assets\u002Fskills\u002Fincident\u002Fpostmortem",1,"https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher","https:\u002F\u002Fgithub.com\u002Fjuchengquan","juchengquan.github.io","A coffee drinker","2026-11-15 01:56:50",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add juchengquan\u002Fkingfisher --skill postmortem",{"tool":38,"command":39},"bunx","bunx skills add juchengquan\u002Fkingfisher --skill postmortem",{"tool":41,"command":42},"pnpm","pnpm dlx skills add juchengquan\u002Fkingfisher --skill postmortem","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fskills\u002Fincident\u002Fpostmortem\u002FSKILL.md",true,[],[47,54,61,68,75,82],{"skillNo":48,"name":49,"nameZh":6,"description":50,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"publishedAt":18},"01M7ANVWQ1W2S8A6A7QNQ3GW91","probe-skill","A skill that exists so seeding has a folder to copy.","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Ftests\u002Funit\u002Fdefinitions\u002Fseed_source\u002Fskills\u002Fprobe-skill\u002FSKILL.md",[],[],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":18},"01M7ANVV1QCFTPAJCQCKYBCV1D","semver","How to report versions once they are ordered, and what to do with a string that is not a version or a tie that precedence cannot break.","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fsubagents\u002Fversions\u002Fskills\u002Fsemver\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":18},"01M7ANVQXW2PC7HQRC8NBEZSEW","iso-8601","How to report a date once it has been read, and what to do with one that cannot be read unambiguously.","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fsubagents\u002Ftimestamps\u002Fskills\u002Fiso-8601\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":18},"01M7ANVMPMQ01Z3GCQ0W5G20GP","tabular-qa","Profiling and quality-checking a delimited data file — row and value counts, duplicate and blank detection, outlier rules, and normalising inconsistent categorical values before counting them.","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fskills\u002Ftabular-qa\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":18},"01M7ANVJDAF8NQMFH27FGCH5Y1","release-notes","Turning a commit range or changelog into release notes written for the people who will read them, grouped by what changed for the user rather than by commit.","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fskills\u002Frelease-notes\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":18},"01M7ANVG2846RX21G7J7NG7P7C","timeline","Building the timeline section of an incident writeup from raw log files — every warning and error in time order, repeats collapsed, silences marked — by running the script this skill ships rather than reading the logs by eye.","https:\u002F\u002Fgithub.com\u002Fjuchengquan\u002Fkingfisher\u002Fblob\u002Fmain\u002Fassets\u002Fskills\u002Fincident\u002Ftimeline\u002FSKILL.md",[],[]]