[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M758TXBEH1282XMYFS2MM5B6":3,"skill-same:01M758TXBEH1282XMYFS2MM5B6":63},{"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},"01M758TXBEH1282XMYFS2MM5B6","playtest-prd-prep",null,"Reconcile one validated Continuity Loom author-playtest report against live repo and tracker state, preserve proven strengths, and write one same-stem change portfolio for playtest-to-issues custody before any later \u002Fto-prd pass. Invoke explicitly with a reports\u002Fplaytest-*.md path; this skill diagnoses and prepares but does not implement or publish.","joeloverbeck\u002Fcontinuity-loom","joeloverbeck","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F120873?v=4","01M758KCBJ0GWAPGN74C9NJF8T","01M758KCCNQ8F0YS327T45F7C0",0,"GPL-3.0","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: playtest-prd-prep\ndescription: Reconcile one validated Continuity Loom author-playtest report against live repo and tracker state, preserve proven strengths, and write one same-stem change portfolio for playtest-to-issues custody before any later \u002Fto-prd pass. Invoke explicitly with a reports\u002Fplaytest-*.md path; this skill diagnoses and prepares but does not implement or publish.\ndisable-model-invocation: true\n---\n\n# Playtest PRD Prep\n\nReconcile a `.claude\u002Fskills\u002Fplaytest` report into one PRD-ready portfolio. The stopping point is\n`reports\u002F\u003Cplaytest-report-stem>-prd-prep.md`. Do not edit the source report, product code, tests,\ndocs, skills, specs, tickets, or tracker state; do not invoke `\u002Fto-prd` or satisfy its seam\ncheckpoint. The closeout must route the completed artifact through `$playtest-to-issues` before\nrecommending `\u002Fto-prd`.\n\n## Input\n\nRequire one canonical `reports\u002Fplaytest-\u003Cstory-slug>-\u003Ctimestamp>.md` path or an unambiguous glob.\nExclude evidence assets and derivative `*-prd-prep.md` files. If the path does not resolve, search\ncase-insensitively under `reports\u002F` and proceed only when one report resolves.\n\nValidate the report and inspect its downstream inventory before trusting it:\n\n```bash\nnode .claude\u002Fskills\u002Fplaytest\u002Fscripts\u002Fvalidate-report.mjs --report \u003Creport>\nnode .claude\u002Fskills\u002Fplaytest-prd-prep\u002Fscripts\u002Fvalidate-prd-prep.mjs --inspect-source \u003Creport>\n```\n\nIf schema, privacy, or ledger errors make exhaustive disposition unsafe, stop without writing or\nupdating the prep artifact. The source inspector may downgrade a bounded prompt-evaluation metadata\nor disclosure mismatch to a non-blocking validator defect when the finding inventory remains\ncomplete; carry that defect into source validation and skill-maintenance follow-up. Identify every\nreport or `playtest`-methodology repair and route skill repair through `$skill-audit`; never repair\nthe source during prep. Missing optional evidence, stale commits, unavailable tracker reads, or\nfailed reproduction do not invalidate the source: continue with explicit limitations and classify\naffected items as verification work.\n\nComplete intake only when the canonical source path, source validation, report inventory, run\nmode, prior-report pointer, prior-report prep state, current same-stem prep state, branch, HEAD,\nand baseline worktree dirt are known.\n\n## Reconcile\n\n### 1. Establish live provenance\n\n1. Retain exact `git status --short --untracked-files=all`, `git branch --show-current`, and\n   `git rev-parse HEAD` output. Treat pre-existing dirt as unowned unless this invocation explicitly\n   adopts it.\n2. Read `AGENTS.md`, `docs\u002FACTIVE-DOCS.md`, `docs\u002FFOUNDATIONS.md`, and the active domain authorities\n   for every affected surface. Read `docs\u002Fagents\u002Fissue-tracker.md`,\n   `docs\u002Fagents\u002Ftriage-labels.md`, and `tickets\u002FREADME.md` before assigning destinations or label\n   posture. Treat `archive\u002F**` only as explicitly routed historical evidence.\n3. Read the complete named report. Use its Cumulative Finding Ledger as the inventory frontier.\n   Follow the explicit `prior_report` chain only for `open`, `repeated`, or `not-retested` rows\n   whose decision-relevant detail is absent from the frontier report. Never scan unrelated reports.\n4. Compare the report's launch HEAD with the current tree over affected paths. A changed HEAD is\n   not itself drift; distinguish relevant committed drift, relevant worktree drift, and unrelated\n   report\u002Fskill drift.\n5. Classify the source report's durability using tracked, clean, publication-ref-visible, and\n   publication-ref-content-identical evidence. Summarize machine-local evidence; never cite its\n   path as durable.\n6. For a continuation with a named `prior_report`, derive that report's same-stem `*-prd-prep.md`\n   path. If it exists, read it as historical recommendation input, never as product authority.\n   Reconcile its former first action, every PRD candidate, and every Non-PRD Follow-Up item as\n   `consumed`, `still live`, `rejected`, or `superseded` against the frontier report, current\n   authorities, implementation, and tracker state. If it is absent, record `missing at intake`.\n7. Check the current report's default same-stem prep path. If it exists, classify it as current,\n   partially consumed, stale, superseded, or not relevant. Reconcile every prior recommendation as\n   `consumed`, `still live`, `rejected`, or `superseded` before reusing it.\n8. Search current open work and relevant closed tracker work from exact report terms and IDs.\n   Start with projected metadata; fetch full bodies or comments only for likely owners. If tracker\n   access is unavailable, record the limitation and do not claim AFK-ready label certainty.\n\nThis step is complete only when every authority and freshness source needed to judge the report is\nchecked or explicitly unavailable, and every recommendation from an existing current same-stem or\nprior-report prep has a current disposition.\n\n### 2. Disposition every report row\n\nGive every Cumulative Finding Ledger row exactly one disposition from the output contract. Current\nPrioritized Findings must already occur in that cumulative ledger; never invent a parallel ID.\n\nTreat report rows as credible observations, not automatic mandates. Promote an observation into a\nstable change only when current source, tests, repeated evidence, deterministic reproduction, or a\ngoverning authority supports the desired rule. Preserve one-run ambiguity and model variability as\nverification, coverage, or research follow-up unless a safe deterministic mitigation is justified.\n\nUse this evidence ladder:\n\n1. report detail and privacy-safe retained evidence;\n2. current authorities, implementation, and focused tests;\n3. exact open\u002Fclosed tracker overlap and closeout evidence;\n4. focused read-only probes; then\n5. a targeted browser reproduction only when drift or contradiction still blocks classification.\n\nFor a targeted browser reproduction, read\n[`..\u002Fplaytest\u002Freferences\u002Fbrowser-driver.md`](..\u002Fplaytest\u002Freferences\u002Fbrowser-driver.md), reuse the\nblank-key guarded localhost boundary, and keep diagnostic state under `\u002Ftmp`. Do not start another\nfull author journey, accept prose, click provider controls, or make OpenRouter requests.\n\nEvery `strength` \u002F `preserve-strength` row is a preservation constraint. Map it to `global` or to\nthe candidates it constrains and name regression evidence. Do not turn a strength into new scope.\n\nThis step is complete only when every source row has one disposition, current evidence, a change\ndestination, and PRD impact; every strength has one preservation row; and no uncategorized blocker\nremains.\n\n### 3. Determine the change portfolio\n\nFollow each report-derived rule across every mechanically affected code, test, active-doc, and\nskill surface. Exclude adjacent issues that do not trace to a report row, preserved strength, or\nrequired consistency consequence.\n\nRoute work to its smallest governed destination:\n\n- broad or risky product behavior becomes a PRD candidate and names any required spec, ADR, active\n  doc, test, or skill changes;\n- narrow, already-scoped product work becomes a ticket candidate;\n- pure playtest-methodology or report-schema work becomes skill maintenance;\n- an active authority that misstates current behavior becomes a doc correction;\n- contradictions with closed work become verification\u002Freopen work;\n- insufficient evidence becomes coverage or research follow-up; and\n- covered, resolved, preference-only, or disproportional work becomes covered or rejected.\n\nBundle PRD rows only when they share the same product rule, decision point, implementation seam,\nand acceptance proof. Default to one recommended first PRD plus deferred follow-ons. Use a\nmulti-PRD program only when independently implementable outcomes have a real dependency or require\none shared ratified direction. It is valid to recommend no new PRD.\n\nRank the first PRD by constitutional\u002Finvariant risk, blocker or author-trust impact, core-loop\nseverity, dependency order, evidence strength, proof readiness, and strength-regression risk.\nKeep `First operational action` separate from the first new PRD when verification, a reopen, or\nskill maintenance must happen first.\n\n`First operational action` is the first substantive portfolio action that custody must resolve,\nnot the mandatory custody or publication handoff. Never set it to `$playtest-to-issues`,\n`playtest-to-issues`, or `\u002Fto-prd`; use `none - \u003Creason>` when no substantive action precedes the\nremaining portfolio.\n\nDescribe behavior, authority, the existing seam, affected surfaces, acceptance, and testing at\nPRD-ready depth. Do not prescribe patch order, volatile symbols, components, or new abstractions\nunless current architecture makes the constraint unavoidable.\n\nIf one material product-scope, sequence, or stewardship choice remains after the evidence ladder,\nask one focused question with a recommendation. Do not conduct a routine requirements interview or\noffload classifications to the user.\n\nThis step is complete only when the portfolio has one publication verdict, one first operational\naction, explicit non-PRD destinations, deferred work, rejected alternatives, and preservation\nconstraints.\n\n### 4. Write the same-stem prep artifact\n\nRead [Prep artifact contract](references\u002Fprep-format.md) in full immediately before writing. Write\nor update the default same-stem artifact; never fork numbered or timestamped copies. On rerun, carry\nthe prior-recommendation consumption ledger into the updated artifact. For a continuation, include\nthe prior report's prep recommendations in that same ledger when its derived prep exists.\n\nAlways write the current `Prep contract version` from the artifact contract. An unversioned or\nolder same-stem prep is historical recommendation input, never a template to preserve byte-for-byte.\nThe producer alone migrates it: inventory its prior recommendations, classify each one against live\nevidence, and rewrite the same-stem artifact at the current version. Do not delegate migration to\n`$playtest-to-issues`.\n\nConsult `\u002Fto-prd` for house style only: publication-package vocabulary, source durability, label\nposture, browser-visible checklist mapping, and recommended testing seams. Do not draft or stage a\nPRD body, create an issue, apply labels, or ask `\u002Fto-prd`'s seam-confirmation question. Record that\nthe seam checkpoint remains owed.\n\nKeep the artifact privacy-safe. Do not copy full prompts, record payloads, raw assistance output,\ncandidate or accepted prose, API keys, localhost URLs, or machine-local paths. Summarize temporary\nevidence and classify it as `summarized, not cited`.\n\nThis step is complete only when `\u002Fto-prd` could select and draft the intended package without\nre-reading the full playtest report to recover provenance, scope, preserved strengths, authority\nimpact, sequencing, or deferred work.\n\n### 5. Validate and close\n\nKeep the `## Completion Self-Check` fields at their draft `pending` values while validating shape.\nRun:\n\n```bash\nnode .claude\u002Fskills\u002Fplaytest-prd-prep\u002Fscripts\u002Fvalidate-prd-prep.mjs --draft \u003Creport> \u003Cprep-artifact>\n```\n\nFix every structural or privacy error. Compare the emitted source and output counts with the report\ntables; the helper proves shape and coverage, never semantic correctness. Manually review each\ndisposition, candidate grouping, authority impact, preservation constraint, and evidence claim,\nthen complete the privacy scan. Immediately before completing the freshness scan, re-run\n`git status --short --untracked-files=all` and `git branch --show-current`; replace the artifact's\nfinal branch and worktree ledger with that exact snapshot and classify every remaining path as the\nintentional prep artifact, pre-existing, or concurrent\u002Funowned. Only after those checks actually\ncomplete, replace the three draft self-check values with the final values from the artifact contract\nand run:\n\n```bash\nnode .claude\u002Fskills\u002Fplaytest-prd-prep\u002Fscripts\u002Fvalidate-prd-prep.mjs \u003Creport> \u003Cprep-artifact>\n```\n\nFix every final-validation error and rerun until it passes. Never leave final completion claims in\nan artifact that has not passed the final command. Immediately after a pass, re-run\n`git status --short --untracked-files=all` and `git branch --show-current` and compare both outputs\nwith the validated snapshot. If either differs, update the snapshot and classifications, repeat the\naffected freshness and privacy review, rerun final validation, and compare again. Stop only when\nthe post-validation comparison matches the validated snapshot exactly. Do not run root gates merely\nfor a report-only prep; report focused tests, probes, browser checks, and broader gates skipped.\n\nThe run is complete only when the source inspector reports `passed` or `nonblocking-defects` with\nthe required skill-maintenance follow-up, every source row and strength is covered, the final prep\nvalidator passes, semantic review is recorded, privacy and freshness scans are clear, and the final\nworktree boundary is reconciled. A closeout that points directly to `\u002Fto-prd` without the exact\n`$playtest-to-issues` custody handoff is incomplete.\n\nBefore sending the final response, render this exact keyed block as the final content from the\nvalidated artifact. Keep every key, use `none` where applicable, and list every remaining path\nindividually under `Remaining dirt` rather than collapsing an untracked directory:\n\n```text\nSource report: \u003Ccanonical source report>\nPrep artifact: \u003Csame-stem prep artifact>\nFirst operational action: \u003Csubstantive action or none - reason>\nPublication verdict: \u003Cpackage and intended first PRD, or no-new-PRD result>\nDeferred work: \u003Cevery deferred PRD candidate, or none>\nNon-PRD work: \u003Cevery item and destination, or none>\nIssue-custody handoff: $playtest-to-issues \"\u003Cprep-artifact>\"; required before \u002Fto-prd\nTracker and durability: \u003Ctracker limitations; source durability; artifact durability>\nIntentional change: \u003Cauthored prep artifact>\nRemaining dirt:\n- \u003Cpath>: \u003Cpre-existing, concurrent\u002Funowned, or intentional classification>\nNon-actions: no implementation, source-report edit, tracker mutation, PRD publication, or \u002Fto-prd seam checkpoint occurred\n```\n","# Playtest PRD Prep\n\nReconcile a `.claude\u002Fskills\u002Fplaytest` report into one PRD-ready portfolio. The stopping point is\n`reports\u002F\u003Cplaytest-report-stem>-prd-prep.md`. Do not edit the source report, product code, tests,\ndocs, skills, specs, tickets, or tracker state; do not invoke `\u002Fto-prd` or satisfy its seam\ncheckpoint. The closeout must route the completed artifact through `$playtest-to-issues` before\nrecommending `\u002Fto-prd`.\n\n## Input\n\nRequire one canonical `reports\u002Fplaytest-\u003Cstory-slug>-\u003Ctimestamp>.md` path or an unambiguous glob.\nExclude evidence assets and derivative `*-prd-prep.md` files. If the path does not resolve, search\ncase-insensitively under `reports\u002F` and proceed only when one report resolves.\n\nValidate the report and inspect its downstream inventory before trusting it:\n\n```bash\nnode .claude\u002Fskills\u002Fplaytest\u002Fscripts\u002Fvalidate-report.mjs --report \u003Creport>\nnode .claude\u002Fskills\u002Fplaytest-prd-prep\u002Fscripts\u002Fvalidate-prd-prep.mjs --inspect-source \u003Creport>\n```\n\nIf schema, privacy, or ledger errors make exhaustive disposition unsafe, stop without writing or\nupdating the prep artifact. The source inspector may downgrade a bounded prompt-evaluation metadata\nor disclosure mismatch to a non-blocking validator defect when the finding inventory remains\ncomplete; carry that defect into source validation and skill-maintenance follow-up. Identify every\nreport or `playtest`-methodology repair and route skill repair through `$skill-audit`; never repair\nthe source during prep. Missing optional evidence, stale commits, unavailable tracker reads, or\nfailed reproduction do not invalidate the source: continue with explicit limitations and classify\naffected items as verification work.\n\nComplete intake only when the canonical source path, source validation, report inventory, run\nmode, prior-report pointer, prior-report prep state, current same-stem prep state, branch, HEAD,\nand baseline worktree dirt are known.\n\n## Reconcile\n\n### 1. Establish live provenance\n\n1. Retain exact `git status --short --untracked-files=all`, `git branch --show-current`, and\n   `git rev-parse HEAD` output. Treat pre-existing dirt as unowned unless this invocation explicitly\n   adopts it.\n2. Read `AGENTS.md`, `docs\u002FACTIVE-DOCS.md`, `docs\u002FFOUNDATIONS.md`, and the active domain authorities\n   for every affected surface. Read `docs\u002Fagents\u002Fissue-tracker.md`,\n   `docs\u002Fagents\u002Ftriage-labels.md`, and `tickets\u002FREADME.md` before assigning destinations or label\n   posture. Treat `archive\u002F**` only as explicitly routed historical evidence.\n3. Read the complete named report. Use its Cumulative Finding Ledger as the inventory frontier.\n   Follow the explicit `prior_report` chain only for `open`, `repeated`, or `not-retested` rows\n   whose decision-relevant detail is absent from the frontier report. Never scan unrelated reports.\n4. Compare the report's launch HEAD with the current tree over affected paths. A changed HEAD is\n   not itself drift; distinguish relevant committed drift, relevant worktree drift, and unrelated\n   report\u002Fskill drift.\n5. Classify the source report's durability using tracked, clean, publication-ref-visible, and\n   publication-ref-content-identical evidence. Summarize machine-local evidence; never cite its\n   path as durable.\n6. For a continuation with a named `prior_report`, derive that report's same-stem `*-prd-prep.md`\n   path. If it exists, read it as historical recommendation input, never as product authority.\n   Reconcile its former first action, every PRD candidate, and every Non-PRD Follow-Up item as\n   `consumed`, `still live`, `rejected`, or `superseded` against the frontier report, current\n   authorities, implementation, and tracker state. If it is absent, record `missing at intake`.\n7. Check the current report's default same-stem prep path. If it exists, classify it as current,\n   partially consumed, stale, superseded, or not relevant. Reconcile every prior recommendation as\n   `consumed`, `still live`, `rejected`, or `superseded` before reusing it.\n8. Search current open work and relevant closed tracker work from exact report terms and IDs.\n   Start with projected metadata; fetch full bodies or comments only for likely owners. If tracker\n   access is unavailable, record the limitation and do not claim AFK-ready label certainty.\n\nThis step is complete only when every authority and freshness source needed to judge the report is\nchecked or explicitly unavailable, and every recommendation from an existing current same-stem or\nprior-report prep has a current disposition.\n\n### 2. Disposition every report row\n\nGive every Cumulative Finding Ledger row exactly one disposition from the output contract. Current\nPrioritized Findings must already occur in that cumulative ledger; never invent a parallel ID.\n\nTreat report rows as credible observations, not automatic mandates. Promote an observation into a\nstable change only when current source, tests, repeated evidence, deterministic reproduction, or a\ngoverning authority supports the desired rule. Preserve one-run ambiguity and model variability as\nverification, coverage, or research follow-up unless a safe deterministic mitigation is justified.\n\nUse this evidence ladder:\n\n1. report detail and privacy-safe retained evidence;\n2. current authorities, implementation, and focused tests;\n3. exact open\u002Fclosed tracker overlap and closeout evidence;\n4. focused read-only probes; then\n5. a targeted browser reproduction only when drift or contradiction still blocks classification.\n\nFor a targeted browser reproduction, read\n[`..\u002Fplaytest\u002Freferences\u002Fbrowser-driver.md`](..\u002Fplaytest\u002Freferences\u002Fbrowser-driver.md), reuse the\nblank-key guarded localhost boundary, and keep diagnostic state under `\u002Ftmp`. Do not start another\nfull author journey, accept prose, click provider controls, or make OpenRouter requests.\n\nEvery `strength` \u002F `preserve-strength` row is a preservation constraint. Map it to `global` or to\nthe candidates it constrains and name regression evidence. Do not turn a strength into new scope.\n\nThis step is complete only when every source row has one disposition, current evidence, a change\ndestination, and PRD impact; every strength has one preservation row; and no uncategorized blocker\nremains.\n\n### 3. Determine the change portfolio\n\nFollow each report-derived rule across every mechanically affected code, test, active-doc, and\nskill surface. Exclude adjacent issues that do not trace to a report row, preserved strength, or\nrequired consistency consequence.\n\nRoute work to its smallest governed destination:\n\n- broad or risky product behavior becomes a PRD candidate and names any required spec, ADR, active\n  doc, test, or skill changes;\n- narrow, already-scoped product work becomes a ticket candidate;\n- pure playtest-methodology or report-schema work becomes skill maintenance;\n- an active authority that misstates current behavior becomes a doc correction;\n- contradictions with closed work become verification\u002Freopen work;\n- insufficient evidence becomes coverage or research follow-up; and\n- covered, resolved, preference-only, or disproportional work becomes covered or rejected.\n\nBundle PRD rows only when they share the same product rule, decision point, implementation seam,\nand acceptance proof. Default to one recommended first PRD plus deferred follow-ons. Use a\nmulti-PRD program only when independently implementable outcomes have a real dependency or require\none shared ratified direction. It is valid to recommend no new PRD.\n\nRank the first PRD by constitutional\u002Finvariant risk, blocker or author-trust impact, core-loop\nseverity, dependency order, evidence strength, proof readiness, and strength-regression risk.\nKeep `First operational action` separate from the first new PRD when verification, a reopen, or\nskill maintenance must happen first.\n\n`First operational action` is the first substantive portfolio action that custody must resolve,\nnot the mandatory custody or publication handoff. Never set it to `$playtest-to-issues`,\n`playtest-to-issues`, or `\u002Fto-prd`; use `none - \u003Creason>` when no substantive action precedes the\nremaining portfolio.\n\nDescribe behavior, authority, the existing seam, affected surfaces, acceptance, and testing at\nPRD-ready depth. Do not prescribe patch order, volatile symbols, components, or new abstractions\nunless current architecture makes the constraint unavoidable.\n\nIf one material product-scope, sequence, or stewardship choice remains after the evidence ladder,\nask one focused question with a recommendation. Do not conduct a routine requirements interview or\noffload classifications to the user.\n\nThis step is complete only when the portfolio has one publication verdict, one first operational\naction, explicit non-PRD destinations, deferred work, rejected alternatives, and preservation\nconstraints.\n\n### 4. Write the same-stem prep artifact\n\nRead [Prep artifact contract](references\u002Fprep-format.md) in full immediately before writing. Write\nor update the default same-stem artifact; never fork numbered or timestamped copies. On rerun, carry\nthe prior-recommendation consumption ledger into the updated artifact. For a continuation, include\nthe prior report's prep recommendations in that same ledger when its derived prep exists.\n\nAlways write the current `Prep contract version` from the artifact contract. An unversioned or\nolder same-stem prep is historical recommendation input, never a template to preserve byte-for-byte.\nThe producer alone migrates it: inventory its prior recommendations, classify each one against live\nevidence, and rewrite the same-stem artifact at the current version. Do not delegate migration to\n`$playtest-to-issues`.\n\nConsult `\u002Fto-prd` for house style only: publication-package vocabulary, source durability, label\nposture, browser-visible checklist mapping, and recommended testing seams. Do not draft or stage a\nPRD body, create an issue, apply labels, or ask `\u002Fto-prd`'s seam-confirmation question. Record that\nthe seam checkpoint remains owed.\n\nKeep the artifact privacy-safe. Do not copy full prompts, record payloads, raw assistance output,\ncandidate or accepted prose, API keys, localhost URLs, or machine-local paths. Summarize temporary\nevidence and classify it as `summarized, not cited`.\n\nThis step is complete only when `\u002Fto-prd` could select and draft the intended package without\nre-reading the full playtest report to recover provenance, scope, preserved strengths, authority\nimpact, sequencing, or deferred work.\n\n### 5. Validate and close\n\nKeep the `## Completion Self-Check` fields at their draft `pending` values while validating shape.\nRun:\n\n```bash\nnode .claude\u002Fskills\u002Fplaytest-prd-prep\u002Fscripts\u002Fvalidate-prd-prep.mjs --draft \u003Creport> \u003Cprep-artifact>\n```\n\nFix every structural or privacy error. Compare the emitted source and output counts with the report\ntables; the helper proves shape and coverage, never semantic correctness. Manually review each\ndisposition, candidate grouping, authority impact, preservation constraint, and evidence claim,\nthen complete the privacy scan. Immediately before completing the freshness scan, re-run\n`git status --short --untracked-files=all` and `git branch --show-current`; replace the artifact's\nfinal branch and worktree ledger with that exact snapshot and classify every remaining path as the\nintentional prep artifact, pre-existing, or concurrent\u002Funowned. Only after those checks actually\ncomplete, replace the three draft self-check values with the final values from the artifact contract\nand run:\n\n```bash\nnode .claude\u002Fskills\u002Fplaytest-prd-prep\u002Fscripts\u002Fvalidate-prd-prep.mjs \u003Creport> \u003Cprep-artifact>\n```\n\nFix every final-validation error and rerun until it passes. Never leave final completion claims in\nan artifact that has not passed the final command. Immediately after a pass, re-run\n`git status --short --untracked-files=all` and `git branch --show-current` and compare both outputs\nwith the validated snapshot. If either differs, update the snapshot and classifications, repeat the\naffected freshness and privacy review, rerun final validation, and compare again. Stop only when\nthe post-validation comparison matches the validated snapshot exactly. Do not run root gates merely\nfor a report-only prep; report focused tests, probes, browser checks, and broader gates skipped.\n\nThe run is complete only when the source inspector reports `passed` or `nonblocking-defects` with\nthe required skill-maintenance follow-up, every source row and strength is covered, the final prep\nvalidator passes, semantic review is recorded, privacy and freshness scans are clear, and the final\nworktree boundary is reconciled. A closeout that points directly to `\u002Fto-prd` without the exact\n`$playtest-to-issues` custody handoff is incomplete.\n\nBefore sending the final response, render this exact keyed block as the final content from the\nvalidated artifact. Keep every key, use `none` where applicable, and list every remaining path\nindividually under `Remaining dirt` rather than collapsing an untracked directory:\n\n```text\nSource report: \u003Ccanonical source report>\nPrep artifact: \u003Csame-stem prep artifact>\nFirst operational action: \u003Csubstantive action or none - reason>\nPublication verdict: \u003Cpackage and intended first PRD, or no-new-PRD result>\nDeferred work: \u003Cevery deferred PRD candidate, or none>\nNon-PRD work: \u003Cevery item and destination, or none>\nIssue-custody handoff: $playtest-to-issues \"\u003Cprep-artifact>\"; required before \u002Fto-prd\nTracker and durability: \u003Ctracker limitations; source durability; artifact durability>\nIntentional change: \u003Cauthored prep artifact>\nRemaining dirt:\n- \u003Cpath>: \u003Cpre-existing, concurrent\u002Funowned, or intentional classification>\nNon-actions: no implementation, source-report edit, tracker mutation, PRD publication, or \u002Fto-prd seam checkpoint occurred\n```\n",{"name":5,"description":7,"disable-model-invocation":23},true,"main","bb13c72ec4134361fec0b0584913aee821330c02","reports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck","2026-11-10 23:00:02",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":23},[32,35,38],{"tool":33,"command":34},"npx","npx skills add joeloverbeck\u002Fcontinuity-loom --skill playtest-prd-prep",{"tool":36,"command":37},"bunx","bunx skills add joeloverbeck\u002Fcontinuity-loom --skill playtest-prd-prep",{"tool":39,"command":40},"pnpm","pnpm dlx skills add joeloverbeck\u002Fcontinuity-loom --skill playtest-prd-prep","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup\u002FSKILL.md",[43,49,53,59],{"path":44,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":47,"sourceHtmlUrl":48},"agents\u002Fopenai.yaml","REFERENCE","参考文档",285,"https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup\u002Fagents\u002Fopenai.yaml",{"path":50,"resourceType":45,"resourceTypeDesc":46,"sizeBytes":51,"sourceHtmlUrl":52},"references\u002Fprep-format.md",17204,"https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup\u002Freferences\u002Fprep-format.md",{"path":54,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":57,"sourceHtmlUrl":58},"scripts\u002Fvalidate-prd-prep.mjs","SCRIPT","脚本",49959,"https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup\u002Fscripts\u002Fvalidate-prd-prep.mjs",{"path":60,"resourceType":55,"resourceTypeDesc":56,"sizeBytes":61,"sourceHtmlUrl":62},"scripts\u002Fvalidate-prd-prep.test.mjs",33128,"https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fplaytest-prd-prep-200fcc29\u002Fdecontamination\u002Fdec_19a21bd7-5ca2-4cb8-a7cb-d2aa701824f4\u002Fpre-land-backup\u002Fscripts\u002Fvalidate-prd-prep.test.mjs",[64,71,76,81,88,93],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":18},"01M758XG3F9AESCQBKYQX0GPSD","triage","Move issues and external PRs through a state machine of triage roles — categorise, verify, grill if needed, and write agent-ready briefs.","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Ftriage\u002Freviews\u002Frev_a4ce832d-b023-4fa0-8d5d-382e558791d8\u002Fcandidate\u002FSKILL.md",[],[],{"skillNo":72,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":18},"01M758XDY39PNPRY8FSXVE8SZN","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Ftriage\u002Fdecontamination\u002Fdec_2d4305f8-ee96-4376-a267-061413c4fdd6\u002Fcandidate\u002FSKILL.md",[],[],{"skillNo":77,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":18},"01M758XB4XZ6N5XP8XZXWVMVQ0","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Ftriage\u002Fdecontamination\u002Fdec_2d4305f8-ee96-4376-a267-061413c4fdd6\u002Fbaseline\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":18},"01M758X7WRCGV9BM5RNEGH4MSX","to-prd","Turn the current conversation into a PRD and publish it to the project issue tracker — no requirements interview, just synthesis of what you've already discussed; the seam confirmation is the sole checkpoint.","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fto-prd\u002Fdecontamination\u002Fdec_581ad8e4-4f9d-4aad-a6ee-59eeec05c8a4\u002Fpre-land-backup\u002FSKILL.md",[],[],{"skillNo":89,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M758X55QJQTVHH1MWHJF2BBZ","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fto-prd\u002Fdecontamination\u002Fdec_581ad8e4-4f9d-4aad-a6ee-59eeec05c8a4\u002Fcandidate\u002FSKILL.md",[],[],{"skillNo":94,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":18},"01M758X1Z0F2YRRHNTR3ZCY02C","https:\u002F\u002Fgithub.com\u002Fjoeloverbeck\u002Fcontinuity-loom\u002Fblob\u002Fmain\u002Freports\u002Fskill-evidence\u002Fto-prd\u002Fdecontamination\u002Fdec_581ad8e4-4f9d-4aad-a6ee-59eeec05c8a4\u002Fbaseline\u002FSKILL.md",[],[]]