[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AHBCTXXVBZTY5BKCDSDRAA":3,"skill-same:01M7AHBCTXXVBZTY5BKCDSDRAA":45},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":14,"repositoryHtmlUrl":29,"repositoryHomepageUrl":6,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":31,"install":32,"resources":44},"01M7AHBCTXXVBZTY5BKCDSDRAA","codex-plan-review",null,"Second-opinion adversarial review of a PLAN file via the Codex CLI, read-only. Use when a plan needs an independent fresh-context review before execution. Manual only. NOT for replacing your project's own plan-review gate; Codex never edits - findings are advisory and Claude folds them into the plan.","SamyakJhaveri\u002Floam","SamyakJhaveri","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F39847642?v=4","01M7AH87180VZDSRHE20JD1S32","Samyak Jhaveri","01M7AH871WNGEHDVNXJM2B520R",0,"MIT","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fcultivation\u002Fparked\u002Fsam-cc-setup\u002Fskills\u002Fcodex-plan-review\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: codex-plan-review\ndisable-model-invocation: true\ndescription: \"Second-opinion adversarial review of a PLAN file via the Codex CLI, read-only. Use when a plan needs an independent fresh-context review before execution. Manual only. NOT for replacing your project's own plan-review gate; Codex never edits - findings are advisory and Claude folds them into the plan.\"\nargument-hint: \"\u003Cpath-to-plan.md>\"\n---\n\n# Codex Plan Review (fresh context)\n\nRun a **plan file** past the **Codex CLI** as an independent fresh-context reviewer and\nfold its findings into the plan. Codex runs in a **read-only**\nsandbox and **never edits the checkout** - it only reports. Claude saves a transcript\nunder `.claude\u002Fcodex-reviews\u002F` (gitignore it) and applies any accepted changes to the plan itself.\n\nThis is the plan-file sibling of `\u002Fcodex-review` (which reviews diffs). Its independence\ncomes from a fresh context, fixed input, and a read-only role. Do not infer review quality\nor independence from an unverified model identity.\n\n**Trigger:** user types `\u002Fcodex-plan-review \u003Cpath>`. Manual-only (`disable-model-invocation:\ntrue`).\n\n## Single-writer rule (mandatory)\n\nCodex must never write to a checkout Claude is using (one checkout = one writer; two\nwriters race on the same working tree). The read-only sandbox enforces this - do\nNOT relax `--sandbox read-only` to `workspace-write` (the repo config default is\nworkspace-write, which would violate this rule). Findings come back as text; **Claude** folds the\naccepted ones into the plan file. Never let Codex update the plan.\n\n## Workflow\n\n### Step 1 - Resolve and validate the plan path\n\n```bash\nPLAN=\"$ARGUMENTS\"      # the plan file to review; no default - a path is required\n\n# The path must be a readable .md file with no options\u002Fmetacharacters (mirror\n# \u002Fcodex-review's validation stance, adapted from a git revision to a file path).\ncase \"$PLAN\" in\n  \"\"|-*)\n    echo \"Usage: \u002Fcodex-plan-review \u003Cpath-to-plan.md> (a readable .md file in the repo).\"\n    exit 2\n    ;;\n  *..*)\n    echo \"Invalid path: parent-directory traversal is not allowed.\"\n    exit 2\n    ;;\n  *[!A-Za-z0-9._\u002F-]*)\n    echo \"Invalid path: spaces, options, and shell metacharacters are not allowed.\"\n    exit 2\n    ;;\n  *.md) ;;\n  *)\n    echo \"Invalid: the plan must be a .md file.\"\n    exit 2\n    ;;\nesac\n\nif [ ! -f \"$PLAN\" ] || [ ! -r \"$PLAN\" ]; then\n  echo \"No readable file at '$PLAN'.\"\n  exit 2\nfi\n\n# Reject an oversized plan: Codex auto-injects AGENTS.md (32 KiB cap) and the plan\n# rides on stdin, so keep the review prompt lean.\nMAX_PLAN_BYTES=150000\nPLAN_BYTES=$(wc -c \u003C \"$PLAN\" | tr -d '[:space:]')\nif [ \"$PLAN_BYTES\" -gt \"$MAX_PLAN_BYTES\" ]; then\n  echo \"Plan is ${PLAN_BYTES} bytes (>150 KB); trim or split it before \u002Fcodex-plan-review.\"\n  exit 2\nfi\n```\n\n### Step 2 - Run Codex (read-only, plan piped via stdin)\n\nPipe the fixed plan on **stdin**. Keep the sandbox read-only. Use the configured Codex\nmodel with high review effort. Inspect `codex --help` and the active provider config before\nadding any model ID or unsupported effort value. `-o` saves the durable final report.\n\n```bash\nPLAN=\"$ARGUMENTS\"   # set explicitly - mirrors codex-review's SCOPE=\"${ARGUMENTS:-...}\" pattern\nmkdir -p .claude\u002Fcodex-reviews\nOUT=\".claude\u002Fcodex-reviews\u002F$(date +%F)-plan-$(basename \"$PLAN\" .md).md\"\nPLAN_FINGERPRINT=$(git hash-object \"$PLAN\")\ncat \"$PLAN\" | codex exec --sandbox read-only \\\n  -c model_reasoning_effort=high -o \"$OUT\" \\\n  \"You are an adversarial plan reviewer. Review the plan in the \u003Cstdin> block against the \\\ncurrent repository (read files read-only for context). Check: (1) every task names real files \\\nand a verification command; (2) unstated assumptions and missing edge cases; (3) ordering\u002F \\\ndependency hazards; (4) ELEGANCE GATE, mandatory and frame-breaking: step back from the plan, \\\nask if a different architecture, built-in feature, existing library, or existing repo machinery \\\nmakes most of the plan unnecessary ('why not just do X instead?'); reuse-before-new is a repo \\\nlaw; a counter-proposal must name what it replaces and its tradeoffs; (5) conformance to \\\n.claude\u002Frules\u002F. \\\nFirst line: verdict APPROVE \u002F APPROVE_WITH_CHANGES \u002F REJECT. Then findings grouped \\\nCritical\u002FHigh\u002FMedium\u002FLow, each with plan-section ref and a one-line concrete fix. Report gaps \\\nthat affect correctness or the stated requirements, not style preferences. \u003C=60 lines.\"\n```\n\n### Step 3 - Fold findings back into the plan (single-writer)\n\n- Read the verdict block (saved at `$OUT`).\n- Treat findings as **advisory**: Claude decides which to accept, then edits the **plan file\n  itself** - Codex does not edit anything.\n- Fold accepted Critical\u002FHigh findings into the plan; note Medium\u002FLow as follow-ups.\n- **Dismiss a finding only with a reproducing check**, never an argument: record the check\n  that shows the finding is a false positive.\n- Record `$PLAN_FINGERPRINT` in the report. Reuse the report while the content and criteria\n  are unchanged. Review again only after material changes or for a named unresolved risk.\n\n## What NOT to do\n\n- Don't run Codex with a writable sandbox (`workspace-write` \u002F `danger-full-access`) - read-only only.\n- Don't let Codex edit the plan - it reviews, Claude folds (single-writer).\n- Not a pipeline gate: `\u002Fcodex-plan-review` complements your project's own plan-review gates, it does not replace them.\n","# Codex Plan Review (fresh context)\n\nRun a **plan file** past the **Codex CLI** as an independent fresh-context reviewer and\nfold its findings into the plan. Codex runs in a **read-only**\nsandbox and **never edits the checkout** - it only reports. Claude saves a transcript\nunder `.claude\u002Fcodex-reviews\u002F` (gitignore it) and applies any accepted changes to the plan itself.\n\nThis is the plan-file sibling of `\u002Fcodex-review` (which reviews diffs). Its independence\ncomes from a fresh context, fixed input, and a read-only role. Do not infer review quality\nor independence from an unverified model identity.\n\n**Trigger:** user types `\u002Fcodex-plan-review \u003Cpath>`. Manual-only (`disable-model-invocation:\ntrue`).\n\n## Single-writer rule (mandatory)\n\nCodex must never write to a checkout Claude is using (one checkout = one writer; two\nwriters race on the same working tree). The read-only sandbox enforces this - do\nNOT relax `--sandbox read-only` to `workspace-write` (the repo config default is\nworkspace-write, which would violate this rule). Findings come back as text; **Claude** folds the\naccepted ones into the plan file. Never let Codex update the plan.\n\n## Workflow\n\n### Step 1 - Resolve and validate the plan path\n\n```bash\nPLAN=\"$ARGUMENTS\"      # the plan file to review; no default - a path is required\n\n# The path must be a readable .md file with no options\u002Fmetacharacters (mirror\n# \u002Fcodex-review's validation stance, adapted from a git revision to a file path).\ncase \"$PLAN\" in\n  \"\"|-*)\n    echo \"Usage: \u002Fcodex-plan-review \u003Cpath-to-plan.md> (a readable .md file in the repo).\"\n    exit 2\n    ;;\n  *..*)\n    echo \"Invalid path: parent-directory traversal is not allowed.\"\n    exit 2\n    ;;\n  *[!A-Za-z0-9._\u002F-]*)\n    echo \"Invalid path: spaces, options, and shell metacharacters are not allowed.\"\n    exit 2\n    ;;\n  *.md) ;;\n  *)\n    echo \"Invalid: the plan must be a .md file.\"\n    exit 2\n    ;;\nesac\n\nif [ ! -f \"$PLAN\" ] || [ ! -r \"$PLAN\" ]; then\n  echo \"No readable file at '$PLAN'.\"\n  exit 2\nfi\n\n# Reject an oversized plan: Codex auto-injects AGENTS.md (32 KiB cap) and the plan\n# rides on stdin, so keep the review prompt lean.\nMAX_PLAN_BYTES=150000\nPLAN_BYTES=$(wc -c \u003C \"$PLAN\" | tr -d '[:space:]')\nif [ \"$PLAN_BYTES\" -gt \"$MAX_PLAN_BYTES\" ]; then\n  echo \"Plan is ${PLAN_BYTES} bytes (>150 KB); trim or split it before \u002Fcodex-plan-review.\"\n  exit 2\nfi\n```\n\n### Step 2 - Run Codex (read-only, plan piped via stdin)\n\nPipe the fixed plan on **stdin**. Keep the sandbox read-only. Use the configured Codex\nmodel with high review effort. Inspect `codex --help` and the active provider config before\nadding any model ID or unsupported effort value. `-o` saves the durable final report.\n\n```bash\nPLAN=\"$ARGUMENTS\"   # set explicitly - mirrors codex-review's SCOPE=\"${ARGUMENTS:-...}\" pattern\nmkdir -p .claude\u002Fcodex-reviews\nOUT=\".claude\u002Fcodex-reviews\u002F$(date +%F)-plan-$(basename \"$PLAN\" .md).md\"\nPLAN_FINGERPRINT=$(git hash-object \"$PLAN\")\ncat \"$PLAN\" | codex exec --sandbox read-only \\\n  -c model_reasoning_effort=high -o \"$OUT\" \\\n  \"You are an adversarial plan reviewer. Review the plan in the \u003Cstdin> block against the \\\ncurrent repository (read files read-only for context). Check: (1) every task names real files \\\nand a verification command; (2) unstated assumptions and missing edge cases; (3) ordering\u002F \\\ndependency hazards; (4) ELEGANCE GATE, mandatory and frame-breaking: step back from the plan, \\\nask if a different architecture, built-in feature, existing library, or existing repo machinery \\\nmakes most of the plan unnecessary ('why not just do X instead?'); reuse-before-new is a repo \\\nlaw; a counter-proposal must name what it replaces and its tradeoffs; (5) conformance to \\\n.claude\u002Frules\u002F. \\\nFirst line: verdict APPROVE \u002F APPROVE_WITH_CHANGES \u002F REJECT. Then findings grouped \\\nCritical\u002FHigh\u002FMedium\u002FLow, each with plan-section ref and a one-line concrete fix. Report gaps \\\nthat affect correctness or the stated requirements, not style preferences. \u003C=60 lines.\"\n```\n\n### Step 3 - Fold findings back into the plan (single-writer)\n\n- Read the verdict block (saved at `$OUT`).\n- Treat findings as **advisory**: Claude decides which to accept, then edits the **plan file\n  itself** - Codex does not edit anything.\n- Fold accepted Critical\u002FHigh findings into the plan; note Medium\u002FLow as follow-ups.\n- **Dismiss a finding only with a reproducing check**, never an argument: record the check\n  that shows the finding is a false positive.\n- Record `$PLAN_FINGERPRINT` in the report. Reuse the report while the content and criteria\n  are unchanged. Review again only after material changes or for a named unresolved risk.\n\n## What NOT to do\n\n- Don't run Codex with a writable sandbox (`workspace-write` \u002F `danger-full-access`) - read-only only.\n- Don't let Codex edit the plan - it reviews, Claude folds (single-writer).\n- Not a pipeline gate: `\u002Fcodex-plan-review` complements your project's own plan-review gates, it does not replace them.\n",{"name":5,"description":7,"argument-hint":24,"disable-model-invocation":25},"\u003Cpath-to-plan.md>",true,"main","85a41de587b98bb440cccf78f081a6bee67fe90e","cultivation\u002Fparked\u002Fsam-cc-setup\u002Fskills\u002Fcodex-plan-review","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri","2026-11-15 04:06:30",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":25},[34,37,40],{"tool":35,"command":36},"npx","npx skills add SamyakJhaveri\u002Floam --skill codex-plan-review",{"tool":38,"command":39},"bunx","bunx skills add SamyakJhaveri\u002Floam --skill codex-plan-review",{"tool":41,"command":42},"pnpm","pnpm dlx skills add SamyakJhaveri\u002Floam --skill codex-plan-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fcultivation\u002Fparked\u002Fsam-cc-setup\u002Fskills\u002Fcodex-plan-review\u002FSKILL.md",[],[46,53,60,67,74,81],{"skillNo":47,"name":48,"nameZh":6,"description":49,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":50,"categories":51,"keywords":52,"publishedAt":19},"01M7AHGVPTGK3JGQ9XYG5XJ6PR","verify-loam","Prove a Loam change in a real rendered project: copier-render this checkout, then drive headless and interactive Claude Code sessions, stubbed Codex runs and copier updates in the render and capture hook, memory, deny-rule, update and check evidence. Use before shipping any seed, hook, settings or copier.yml change.","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fseed\u002F.claude\u002Fskills\u002Fverify-loam\u002FSKILL.md",[],[],{"skillNo":54,"name":55,"nameZh":6,"description":56,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":57,"categories":58,"keywords":59,"publishedAt":19},"01M7AHGEDXX7GWJ32S9BZAEWAJ","hypothesis-tree","Persistent hypothesis tree for multi-step, multi-session investigations. Use when a \"why does X fail\" question branches into sub-hypotheses that must survive across sessions, or when falsifiable claims need tracking with dated, linked evidence. Maintains one markdown file with add, update, review, and prune subcommands, each gated on verifying the cited evidence file exists and supports the claim. NOT for a single-session bug hunt, a one-command question, or general note-taking.","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fseed\u002F.agents\u002Fskills\u002Fhypothesis-tree\u002FSKILL.md",[],[],{"skillNo":61,"name":62,"nameZh":6,"description":63,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M7AHGADA1CX9M0B0QSFJYGW5","fable-prompting","Which Claude Fable 5.2 guide sections you can act on from a session, and which Claude Code already injects. Use before writing a Fable session prompt or brief.","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fseed\u002F.agents\u002Fskills\u002Ffable-prompting\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":70,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":71,"categories":72,"keywords":73,"publishedAt":19},"01M7AHG7AGBSQ6HH04C70TSM8H","catchup","Read-only session bootstrap briefing of git state, commits, handoff, tasks and red flags. Use when resuming work, starting a fresh session, or unsure of project state.","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fseed\u002F.agents\u002Fskills\u002Fcatchup\u002FSKILL.md",[],[],{"skillNo":75,"name":76,"nameZh":6,"description":77,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":78,"categories":79,"keywords":80,"publishedAt":19},"01M7AHG52XX2N7AT7JNHS6MN1S","writing-for-agents","Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fdocs\u002Farchive\u002Farchitecture-working\u002Ftooling\u002Fmattpocock-skills\u002Fwriting-for-agents\u002FSKILL.md",[],[],{"skillNo":82,"name":83,"nameZh":6,"description":84,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":85,"categories":86,"keywords":87,"publishedAt":19},"01M7AHG00SXGAGJA7C6XA193J6","wizard","Generate an interactive bash wizard that walks a human through steps only they can perform. Use when provisioning infrastructure, setting up credentials or CI secrets, walking an unfamiliar third-party dashboard, or running a one-off migration or cutover. Don't invoke this for steps the agent can perform itself.","https:\u002F\u002Fgithub.com\u002FSamyakJhaveri\u002Floam\u002Fblob\u002Fmain\u002Fdocs\u002Farchive\u002Farchitecture-working\u002Ftooling\u002Fmattpocock-skills\u002Fwizard\u002FSKILL.md",[],[]]