[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78F29W607Y8S0NJ7YDR2RF7":3,"skill-same:01M78F29W607Y8S0NJ7YDR2RF7":47},{"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":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M78F29W607Y8S0NJ7YDR2RF7","checkpoint",null,"Use when ending a work session, pausing, hitting a context limit, or before any handoff - captures a canonical stopping point so no context is lost. Also use when resuming, to recover the previous session. Replaces hand-written devlog\u002FRESUME prose with typed, staleness-aware records.","CharlesHoskinson\u002Fforeman","CharlesHoskinson","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F3279952?v=4","01M78F029R4HEDS222354N3VAX","Charles Hoskinson","01M78F02AN4AQKGVVK97YEG5NQ",9,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fcheckpoint\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: checkpoint\ndescription: Use when ending a work session, pausing, hitting a context limit, or before any handoff - captures a canonical stopping point so no context is lost. Also use when resuming, to recover the previous session. Replaces hand-written devlog\u002FRESUME prose with typed, staleness-aware records.\n---\n\n# checkpoint — canonical session stopping point\n\nA checkpoint is not a summary. A summary is prose someone must read and\nreconcile; this is **typed records a machine can hand back verbatim**, with\nevery perishable claim carrying a computed freshness verdict.\n\nThe failure this exists to prevent, observed repeatedly: an agent resumes, reads\na hand-written RESUME doc, and confidently quotes \"13\u002F13 tests green\" that was\nmeasured hours and eleven commits earlier. The number was never wrong when\nwritten. It went stale, and flat prose cannot say so.\n\nStore: `.foreman\u002Fsession.db` (SQLite).\nTool: `skills\u002Fforeman\u002Fscripts\u002Ffm-session.py`.\n\n## Resuming — do this FIRST, before anything else\n\n```bash\npython3 skills\u002Fforeman\u002Fscripts\u002Ffm-session.py begin --note \"what you are about to do\"\n```\n\nThis prints the canonical recovery and mints a session id. It is the **only**\nrecovery you need — do not also go reading old RESUME files, and do not trust a\nnumber in any markdown file over what this prints.\n\nRead the output in this order:\n\n1. **STALE \u002F unknown measurements.** Each carries the exact command that\n   produced it. **Re-run them before quoting any of their numbers.** This is the\n   whole point of the checkpoint system; skipping it reintroduces the bug.\n2. **Open obligations.** This is your work queue.\n3. **Facts.** Durable context — what landed, with evidence.\n\n## Creating a checkpoint — when to do it\n\nDo this when the user says checkpoint, when you are pausing or handing off,\n**when you sense the context window filling**, and before any long-running\ndispatch whose result you would otherwise have to remember.\n\nCapture in this order. Everything is `python3 skills\u002Fforeman\u002Fscripts\u002Ffm-session.py …`.\n\n### 1. Facts — durable, true by construction\n\nThings that stay true: a commit landed, a decision was made, a package shipped.\nAlways attach evidence (a SHA, a path, a command).\n\n```bash\nfm-session.py fact \"vendor-preflight fixed: auth binds to probe content not exit status\" --evidence \"commit ad2366f\"\nfm-session.py fact \"SQLite ontology replaced TerminusDB\" --evidence \"commit b3bbdc3\"\n```\n\n### 2. Measurements — perishable, and the reason this system exists\n\nAny number produced by running something: test counts, timings, benchmark\nresults, pass\u002Ffail verdicts.\n\n**`--scope` is mandatory and the tool refuses without it.** The scope is the set\nof paths whose modification would invalidate the number. A measurement with no\nscope can never be shown stale, which defeats the entire mechanism.\n\n```bash\nfm-session.py measure \"tests\u002Faudit-verdict.bats pass count\" 26 \\\n  --command \"flock \u002Ftmp\u002Fforeman-bats.lock bats tests\u002Faudit-verdict.bats\" \\\n  --scope tests\u002Faudit-verdict.bats \\\n  --scope skills\u002Fforeman\u002Fscripts\u002Faudit-run.sh\n```\n\nRecovery then computes `fresh|stale|unknown` at read time via\n`git rev-list \u003Cmeasured_sha>..HEAD -- \u003Cscope>`. You cannot receive the number\nwithout its verdict.\n\n**Never record process liveness as a measurement.** \"Lane X is running\" is false\nseconds later. Ask the process; that is what `tools\u002Flanectl.sh ps` is for.\n\n### 3. Obligations — owed work and stated blockers\n\n```bash\nfm-session.py obligation \"wt-merge.sh gate-to-merge TOCTOU needs its own package\"\nfm-session.py obligation \"graph-context-builder T2-T10 unimplemented\" --blocker \"needs architect diagnosis; task lists do not dispatch\"\nfm-session.py close 3 --status done\n```\n\nA stated blocker is a good outcome. Record it rather than leaving it implied.\n\n### 4. Supersede rather than edit\n\nWhen a fact stops being true, do not delete it — link the replacement, so the\nhistory of what was believed stays intact:\n\n```bash\nfm-session.py supersede 4 \"the suite is 41 files, not 33\" --evidence \"tests\u002Frun.sh output\"\n```\n\n### 5. End the session\n\n```bash\nfm-session.py end\n```\n\n## What belongs here versus elsewhere\n\n| Goes in the checkpoint | Stays where it is |\n|---|---|\n| What landed, with evidence | Full narrative devlog prose |\n| Numbers, with the command and scope that produced them | `AGENT_TRAPS.md` — standing rules that never go stale |\n| Owed work and blockers | `bugeventlog.md` — incident post-mortems |\n| Decisions and their reasons | Source code and its comments |\n\nThe checkpoint answers *\"what is true right now, and what of it can I still\ntrust?\"* The other files answer *\"what happened\"* and *\"what must I never do\nagain.\"* Do not merge them.\n\n## Rules\n\n- **Recovery is exact SQL, never similarity search.** Two resumes of the same\n  tree must produce the same world. If recovery ever becomes fuzzy, it is no\n  longer canonical and you are back to guessing.\n- **A measurement without `--scope` is refused.** Do not work around it by\n  recording numbers as facts — a fact cannot go stale, so that launders a\n  perishable claim into a durable one. That is precisely the original bug.\n- **Re-run stale measurements before quoting them.** The recovery output hands\n  you the command; there is no excuse.\n- Checkpoint *before* the context window is critical, not after. A checkpoint\n  you could not finish writing is worth nothing.\n","# checkpoint — canonical session stopping point\n\nA checkpoint is not a summary. A summary is prose someone must read and\nreconcile; this is **typed records a machine can hand back verbatim**, with\nevery perishable claim carrying a computed freshness verdict.\n\nThe failure this exists to prevent, observed repeatedly: an agent resumes, reads\na hand-written RESUME doc, and confidently quotes \"13\u002F13 tests green\" that was\nmeasured hours and eleven commits earlier. The number was never wrong when\nwritten. It went stale, and flat prose cannot say so.\n\nStore: `.foreman\u002Fsession.db` (SQLite).\nTool: `skills\u002Fforeman\u002Fscripts\u002Ffm-session.py`.\n\n## Resuming — do this FIRST, before anything else\n\n```bash\npython3 skills\u002Fforeman\u002Fscripts\u002Ffm-session.py begin --note \"what you are about to do\"\n```\n\nThis prints the canonical recovery and mints a session id. It is the **only**\nrecovery you need — do not also go reading old RESUME files, and do not trust a\nnumber in any markdown file over what this prints.\n\nRead the output in this order:\n\n1. **STALE \u002F unknown measurements.** Each carries the exact command that\n   produced it. **Re-run them before quoting any of their numbers.** This is the\n   whole point of the checkpoint system; skipping it reintroduces the bug.\n2. **Open obligations.** This is your work queue.\n3. **Facts.** Durable context — what landed, with evidence.\n\n## Creating a checkpoint — when to do it\n\nDo this when the user says checkpoint, when you are pausing or handing off,\n**when you sense the context window filling**, and before any long-running\ndispatch whose result you would otherwise have to remember.\n\nCapture in this order. Everything is `python3 skills\u002Fforeman\u002Fscripts\u002Ffm-session.py …`.\n\n### 1. Facts — durable, true by construction\n\nThings that stay true: a commit landed, a decision was made, a package shipped.\nAlways attach evidence (a SHA, a path, a command).\n\n```bash\nfm-session.py fact \"vendor-preflight fixed: auth binds to probe content not exit status\" --evidence \"commit ad2366f\"\nfm-session.py fact \"SQLite ontology replaced TerminusDB\" --evidence \"commit b3bbdc3\"\n```\n\n### 2. Measurements — perishable, and the reason this system exists\n\nAny number produced by running something: test counts, timings, benchmark\nresults, pass\u002Ffail verdicts.\n\n**`--scope` is mandatory and the tool refuses without it.** The scope is the set\nof paths whose modification would invalidate the number. A measurement with no\nscope can never be shown stale, which defeats the entire mechanism.\n\n```bash\nfm-session.py measure \"tests\u002Faudit-verdict.bats pass count\" 26 \\\n  --command \"flock \u002Ftmp\u002Fforeman-bats.lock bats tests\u002Faudit-verdict.bats\" \\\n  --scope tests\u002Faudit-verdict.bats \\\n  --scope skills\u002Fforeman\u002Fscripts\u002Faudit-run.sh\n```\n\nRecovery then computes `fresh|stale|unknown` at read time via\n`git rev-list \u003Cmeasured_sha>..HEAD -- \u003Cscope>`. You cannot receive the number\nwithout its verdict.\n\n**Never record process liveness as a measurement.** \"Lane X is running\" is false\nseconds later. Ask the process; that is what `tools\u002Flanectl.sh ps` is for.\n\n### 3. Obligations — owed work and stated blockers\n\n```bash\nfm-session.py obligation \"wt-merge.sh gate-to-merge TOCTOU needs its own package\"\nfm-session.py obligation \"graph-context-builder T2-T10 unimplemented\" --blocker \"needs architect diagnosis; task lists do not dispatch\"\nfm-session.py close 3 --status done\n```\n\nA stated blocker is a good outcome. Record it rather than leaving it implied.\n\n### 4. Supersede rather than edit\n\nWhen a fact stops being true, do not delete it — link the replacement, so the\nhistory of what was believed stays intact:\n\n```bash\nfm-session.py supersede 4 \"the suite is 41 files, not 33\" --evidence \"tests\u002Frun.sh output\"\n```\n\n### 5. End the session\n\n```bash\nfm-session.py end\n```\n\n## What belongs here versus elsewhere\n\n| Goes in the checkpoint | Stays where it is |\n|---|---|\n| What landed, with evidence | Full narrative devlog prose |\n| Numbers, with the command and scope that produced them | `AGENT_TRAPS.md` — standing rules that never go stale |\n| Owed work and blockers | `bugeventlog.md` — incident post-mortems |\n| Decisions and their reasons | Source code and its comments |\n\nThe checkpoint answers *\"what is true right now, and what of it can I still\ntrust?\"* The other files answer *\"what happened\"* and *\"what must I never do\nagain.\"* Do not merge them.\n\n## Rules\n\n- **Recovery is exact SQL, never similarity search.** Two resumes of the same\n  tree must produce the same world. If recovery ever becomes fuzzy, it is no\n  longer canonical and you are back to guessing.\n- **A measurement without `--scope` is refused.** Do not work around it by\n  recording numbers as facts — a fact cannot go stale, so that launders a\n  perishable claim into a durable one. That is precisely the original bug.\n- **Re-run stale measurements before quoting them.** The recovery output hands\n  you the command; there is no excuse.\n- Checkpoint *before* the context window is critical, not after. A checkpoint\n  you could not finish writing is worth nothing.\n",{"name":5,"description":7},"main","58f1b25a52305f5b14eba143e37a6b30466e9bf8","skills\u002Fcheckpoint",4,"https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson","https:\u002F\u002Fwww.iog.io\u002F","CEO of IOG","2026-09-16 08:17:40",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add CharlesHoskinson\u002Fforeman --skill checkpoint",{"tool":39,"command":40},"bunx","bunx skills add CharlesHoskinson\u002Fforeman --skill checkpoint",{"tool":42,"command":43},"pnpm","pnpm dlx skills add CharlesHoskinson\u002Fforeman --skill checkpoint","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fcheckpoint\u002FSKILL.md",true,[],[48,54,61,68,75,82],{"skillNo":49,"name":50,"nameZh":6,"description":6,"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},"01M78F4F3YJ7YN8V4JN5EF4SNE","foreman","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Ftool-check\u002Fcheck-one\u002Fhome-ok\u002F.claude\u002Fskills\u002Fforeman\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},"01M78F4AAGHH448E8CBSSMVZBF","writing-skills","Use when creating new skills, editing existing skills, or verifying skills work before deployment","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers\u002Fskills\u002Fwriting-skills\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},"01M78F4846T6VZB3WZVBBFZB6A","writing-plans","Use when you have a spec or requirements for a multi-step task, before touching code","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers\u002Fskills\u002Fwriting-plans\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},"01M78F45C2CQ2XPQ4SVQF6STY0","verification-before-completion","Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers\u002Fskills\u002Fverification-before-completion\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},"01M78F439T6S7EME6XMB7MG5CV","using-superpowers","Use when starting any conversation - establishes how to find and use skills, requiring skill invocation before ANY response including clarifying questions","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers\u002Fskills\u002Fusing-superpowers\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},"01M78F3ZZ8QFXAFBK5ZBMW1S9F","using-git-worktrees","Use when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback","https:\u002F\u002Fgithub.com\u002FCharlesHoskinson\u002Fforeman\u002Fblob\u002Fmain\u002Fskills\u002Fsuperpowers\u002Fskills\u002Fusing-git-worktrees\u002FSKILL.md",[],[]]