[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79H6XC3ZB07MKFWTM4FFAXK":3,"skill-same:01M79H6XC3ZB07MKFWTM4FFAXK":46},{"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":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":29,"ownerDescription":30,"updatedAt":31,"install":32,"resources":45},"01M79H6XC3ZB07MKFWTM4FFAXK","preflight",null,"Verify a room's recorded work state against Actuality and Canon before taking a big vertical slice — every ticket diagnosed, every block tested for whether it is real. Use before a slice, an integration advance, or a Captain dispatch, and whenever recorded state looks older than the work.","KaydenClark\u002FFoundry","KaydenClark","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F26465021?v=4","01M79H31FNE2HKKE40PPQA5VCC","Kayden Clark","01M79H31GPGM16FRE9M4QXJT4T",0,"MIT","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Fflight-skills\u002Fpreflight\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: preflight\ndescription: Verify a room's recorded work state against Actuality and Canon before taking a big vertical slice — every ticket diagnosed, every block tested for whether it is real. Use before a slice, an integration advance, or a Captain dispatch, and whenever recorded state looks older than the work.\n---\n\n# Preflight\n\nAnswer, for every ticket in scope: **is this really blocked, and do we have\nwhat we need to finish it?** Then either authorize the launch or name exactly\nwhat must be reconciled first.\n\nA sitrep is a **read of Projection** — the top of the stack, a capture of what\nstate *was* when someone last looked. A preflight goes down to **Actuality**\nand **Canon**: the live filesystem, the real refs, the commits a ticket cites\nas proof. That is the whole difference. A sitrep may route; only a passed\npreflight authorizes execution.\n\n## Run it\n\nUse ticket-wide audit only to diagnose recorded state:\n\n```bash\nnode tools\u002Fpreflight.mjs --audit --repo \u002FABSOLUTE\u002FREPO [--repo \u002FABSOLUTE\u002FREPO2]\n```\n\nScope to one spec with `--spec S-###`; omit it to audit every active spec.\nPass `--repo` for **every repository the work touches** — a room's tickets\nroutinely land in a sibling repo, and an unscanned repo is an invisible one.\nAdd `--json` when you need the structured result. `--audit` remains diagnostic\nand non-authorizing: it never selects or authorizes a Job Order launch.\n\nTo gate one specific launch instead of diagnosing all of them:\n\n```bash\nnode tools\u002Fpreflight.mjs --spec S-### --ticket TK-### --job-order JO-XXXXXX --repo \u002FABSOLUTE\u002FREPO [--repo \u002FABSOLUTE\u002FREPO2] [--push-to BRANCH] --json\n```\n\nLaunch ingress requires all three matching selectors: exact Spec, ticket, and\nJob Order alias. A ticket-only launch returns `job-order-required` and is\nnon-authorizing. Do not infer a Job Order from a ticket, prose, branch name,\ndirectory, ordering, or the first apparently ready packet.\n\nAn exit code of zero is necessary but not sufficient. Read the structured\nresult and confirm `jobOrder.alias`, `jobOrder.revision`, and\n`launchAuthorized` exactly match the Canon-issued alias, revision, and literal\n`true`. A missing, different, stale, ambiguous, or ineligible order fails the\nlaunch even if another order under the ticket could pass.\n\n## Read the exit code before anything else\n\n| Exit | Meaning | What you do |\n|---|---|---|\n| 0 | Every check passed | Launch is authorized only when the structured exact-order checks above also match |\n| 3 | **Reconcilable only** — the record disagrees with source | Go verify against Actuality and update the record. **Do not stop and wait for the owner.** |\n| 1 | Blocking | Stop. This is an owner report. |\n| 2 | Usage error | Fix the invocation |\n\nExit 3 is the point of this skill. A stale record and a live owner gate used\nto produce the same halt, so a lane whose paperwork had simply fallen behind\nwould sit waiting on Kayden for hours. They are now different outcomes with\ndifferent responses.\n\n## Per-ticket verdicts\n\n| Verdict | Means |\n|---|---|\n| `launchable` | Blockers satisfied and the Next gate authorizes it |\n| `gate-elsewhere` | Nothing holds it, but the Next gate names another ticket |\n| `blocked-verified` | Held by a named ticket that is genuinely not done |\n| `blocked-stale` | Held by a ticket that is already finished — **reconcile, do not wait** |\n| `blocked-unverifiable` | Held by prose no tool can check; a human must read it |\n| `owner-gated` | Reserved to Kayden; only an Owner Command clears it |\n| `proof-dangling` | Recorded done, but its cited proof does not resolve anywhere |\n| `deferred` | A decision, not an obstacle |\n| `done` | Closed; says whether the proof actually resolves |\n| `unevaluable` | No reachable evidence either way — fails closed, treat as unknown |\n\n## What to do with the findings\n\n1. **Reconcile before you launch, not after.** For each `blocked-stale`,\n   `proof-dangling`, or stale-ref finding: open the branch or commit it names,\n   confirm what actually landed, then correct the ticket row. Reconciling a\n   record to match verified source is inside any agent's authority; it is\n   bookkeeping, not a scope change.\n2. **Never release an owner gate to clear a finding.** `owner-gated` and every\n   blocking failure stay with Kayden. Reconciling the record is not the same\n   act as releasing a guard someone deliberately placed — if you cannot tell\n   which one you are doing, it is the second.\n3. **Re-run after reconciling.** The gate is reconciled, never bypassed. A\n   corrected record that still fails is a real failure.\n4. **Say what you could not check.** `unevaluable`, `blocked-unverifiable`,\n   and a ticket citing no machine-checkable proof are all reported as unknown\n   rather than assumed fine. Repeat that honestly instead of rounding up.\n\n## Limits worth stating out loud\n\nThe tool verifies what the record *cites*. A ticket whose Proof cell is empty\noffers nothing to check, so a lane can be stale in a way no check will catch —\nwhich is exactly how S-024 stalled. Cite commits and paths in Proof cells and\nthis stays detectable; leave them blank and it does not.\n\nIt also matches refs by spec id or by branches the spec names in its own\nprose. A bare ticket id is deliberately not matched: nearly every spec has a\nTK-005, and matching it alone reported one branch against fifteen specs.\n\nPreflight is read-only and mutates nothing. It creates no durable artifact —\nif reconciliation produces knowledge worth keeping, that is a separate `\u002Fsave`\nor `\u002Fto-docs` step under its own authority.\n\nEvery rejection is read-only and causes no mutation. A passed Preflight does\nnot create a Claim, Run, receipt, or lifecycle identity; it is only the exact,\nfresh launch gate consumed by the separately authorized next stage.\n","# Preflight\n\nAnswer, for every ticket in scope: **is this really blocked, and do we have\nwhat we need to finish it?** Then either authorize the launch or name exactly\nwhat must be reconciled first.\n\nA sitrep is a **read of Projection** — the top of the stack, a capture of what\nstate *was* when someone last looked. A preflight goes down to **Actuality**\nand **Canon**: the live filesystem, the real refs, the commits a ticket cites\nas proof. That is the whole difference. A sitrep may route; only a passed\npreflight authorizes execution.\n\n## Run it\n\nUse ticket-wide audit only to diagnose recorded state:\n\n```bash\nnode tools\u002Fpreflight.mjs --audit --repo \u002FABSOLUTE\u002FREPO [--repo \u002FABSOLUTE\u002FREPO2]\n```\n\nScope to one spec with `--spec S-###`; omit it to audit every active spec.\nPass `--repo` for **every repository the work touches** — a room's tickets\nroutinely land in a sibling repo, and an unscanned repo is an invisible one.\nAdd `--json` when you need the structured result. `--audit` remains diagnostic\nand non-authorizing: it never selects or authorizes a Job Order launch.\n\nTo gate one specific launch instead of diagnosing all of them:\n\n```bash\nnode tools\u002Fpreflight.mjs --spec S-### --ticket TK-### --job-order JO-XXXXXX --repo \u002FABSOLUTE\u002FREPO [--repo \u002FABSOLUTE\u002FREPO2] [--push-to BRANCH] --json\n```\n\nLaunch ingress requires all three matching selectors: exact Spec, ticket, and\nJob Order alias. A ticket-only launch returns `job-order-required` and is\nnon-authorizing. Do not infer a Job Order from a ticket, prose, branch name,\ndirectory, ordering, or the first apparently ready packet.\n\nAn exit code of zero is necessary but not sufficient. Read the structured\nresult and confirm `jobOrder.alias`, `jobOrder.revision`, and\n`launchAuthorized` exactly match the Canon-issued alias, revision, and literal\n`true`. A missing, different, stale, ambiguous, or ineligible order fails the\nlaunch even if another order under the ticket could pass.\n\n## Read the exit code before anything else\n\n| Exit | Meaning | What you do |\n|---|---|---|\n| 0 | Every check passed | Launch is authorized only when the structured exact-order checks above also match |\n| 3 | **Reconcilable only** — the record disagrees with source | Go verify against Actuality and update the record. **Do not stop and wait for the owner.** |\n| 1 | Blocking | Stop. This is an owner report. |\n| 2 | Usage error | Fix the invocation |\n\nExit 3 is the point of this skill. A stale record and a live owner gate used\nto produce the same halt, so a lane whose paperwork had simply fallen behind\nwould sit waiting on Kayden for hours. They are now different outcomes with\ndifferent responses.\n\n## Per-ticket verdicts\n\n| Verdict | Means |\n|---|---|\n| `launchable` | Blockers satisfied and the Next gate authorizes it |\n| `gate-elsewhere` | Nothing holds it, but the Next gate names another ticket |\n| `blocked-verified` | Held by a named ticket that is genuinely not done |\n| `blocked-stale` | Held by a ticket that is already finished — **reconcile, do not wait** |\n| `blocked-unverifiable` | Held by prose no tool can check; a human must read it |\n| `owner-gated` | Reserved to Kayden; only an Owner Command clears it |\n| `proof-dangling` | Recorded done, but its cited proof does not resolve anywhere |\n| `deferred` | A decision, not an obstacle |\n| `done` | Closed; says whether the proof actually resolves |\n| `unevaluable` | No reachable evidence either way — fails closed, treat as unknown |\n\n## What to do with the findings\n\n1. **Reconcile before you launch, not after.** For each `blocked-stale`,\n   `proof-dangling`, or stale-ref finding: open the branch or commit it names,\n   confirm what actually landed, then correct the ticket row. Reconciling a\n   record to match verified source is inside any agent's authority; it is\n   bookkeeping, not a scope change.\n2. **Never release an owner gate to clear a finding.** `owner-gated` and every\n   blocking failure stay with Kayden. Reconciling the record is not the same\n   act as releasing a guard someone deliberately placed — if you cannot tell\n   which one you are doing, it is the second.\n3. **Re-run after reconciling.** The gate is reconciled, never bypassed. A\n   corrected record that still fails is a real failure.\n4. **Say what you could not check.** `unevaluable`, `blocked-unverifiable`,\n   and a ticket citing no machine-checkable proof are all reported as unknown\n   rather than assumed fine. Repeat that honestly instead of rounding up.\n\n## Limits worth stating out loud\n\nThe tool verifies what the record *cites*. A ticket whose Proof cell is empty\noffers nothing to check, so a lane can be stale in a way no check will catch —\nwhich is exactly how S-024 stalled. Cite commits and paths in Proof cells and\nthis stays detectable; leave them blank and it does not.\n\nIt also matches refs by spec id or by branches the spec names in its own\nprose. A bare ticket id is deliberately not matched: nearly every spec has a\nTK-005, and matching it alone reported one branch against fifteen specs.\n\nPreflight is read-only and mutates nothing. It creates no durable artifact —\nif reconciliation produces knowledge worth keeping, that is a separate `\u002Fsave`\nor `\u002Fto-docs` step under its own authority.\n\nEvery rejection is read-only and causes no mutation. A passed Preflight does\nnot create a Claim, Run, receipt, or lifecycle identity; it is only the exact,\nfresh launch gate consumed by the separately authorized next stage.\n",{"name":5,"description":7},"main","9381af33e9db4f69f7c4f83ded287e23e109c7c1","recovery\u002Fflight-skills\u002Fpreflight","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry","https:\u002F\u002Fgithub.com\u002FKaydenClark","https:\u002F\u002Fkayden-clark.vercel.app\u002F","Building simple systems to solve complex problems","2026-09-05 15:28:09",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add KaydenClark\u002FFoundry --skill preflight",{"tool":38,"command":39},"bunx","bunx skills add KaydenClark\u002FFoundry --skill preflight",{"tool":41,"command":42},"pnpm","pnpm dlx skills add KaydenClark\u002FFoundry --skill preflight","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Fflight-skills\u002Fpreflight\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":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"publishedAt":19},"01M79H8BQMY5ZFKYHRQ12YT67B","writing-great-skills","Reference for writing and editing skills well — the vocabulary and principles that make a skill predictable.","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Flegacy-skill-snapshot\u002Fwriting-great-skills\u002FSKILL.md",[],[],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M79H88YTKNDMAPEHS9HDM51E","update-harness","Update a project to the current canonical LLM Workbench harness without losing project-specific controls, work state, or proof.","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Flegacy-skill-snapshot\u002Fupdate-harness\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":19},"01M79H867VMC0SQ5A1687SCGCF","tracer-bullet","Cut a capability into vertical tracer-bullet slices that each pierce every layer of a project's stack, then order, scope, and assign them.","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Flegacy-skill-snapshot\u002Ftracer-bullet\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M79H84D6KN135D17505TFEXH","to-tickets","Break an assigned Workbench spec into dependency-aware tracer-bullet tickets.","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Flegacy-skill-snapshot\u002Fto-tickets\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M79H82Q2GFBBCQZ4ZHHHYDW1","to-spec","Turn an already-settled conversation into one stable Workbench capability spec without restarting the interview.","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Flegacy-skill-snapshot\u002Fto-spec\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M79H80SAT2SWTSF8HA4J8X34","to-docs","Route settled conversation truth into existing Workbench documentation owners without restarting discovery or creating another store.","https:\u002F\u002Fgithub.com\u002FKaydenClark\u002FFoundry\u002Fblob\u002Fmain\u002Frecovery\u002Flegacy-skill-snapshot\u002Fto-docs\u002FSKILL.md",[],[]]