[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C8N2G2Q3RW3TPQGATGT682":3,"skill-same:01M7C8N2G2Q3RW3TPQGATGT682":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},"01M7C8N2G2Q3RW3TPQGATGT682","codex",null,"Use the local Codex CLI as an independent second agent. Two branches — (1) proactively run `codex review` for a second opinion after completing a substantive change, before presenting it as done or committing; (2) delegate a well-defined implementation task via `codex exec`, ONLY when the user explicitly asks for Codex to do it. Also covers how to prompt Codex.","transitive-bullshit\u002Fskills","transitive-bullshit","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F552829?v=4","01M7C8KS4F4572TZFKR12N2A3W","Travis Fischer","01M7C8KS56YA2KDQHDNXEGS23P",4,"MIT","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fprovenance\u002Fupstream\u002Fcodex\u002Fskill\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: codex\ndescription: Use the local Codex CLI as an independent second agent. Two branches — (1) proactively run `codex review` for a second opinion after completing a substantive change, before presenting it as done or committing; (2) delegate a well-defined implementation task via `codex exec`, ONLY when the user explicitly asks for Codex to do it. Also covers how to prompt Codex.\n---\n\n# Codex\n\nCodex is an independent agent on PATH (`codex` — invoke as `command codex` if\na shell alias shadows it), sharing this working tree and already authenticated. It is a second opinion, not ground truth: verify what it\nreports, own what it changes. It reads the same skills your repo carries.\n\nLeave Codex's model at its CLI default unless the user asks for a specific\none. The default follows the current Codex release; do not pin a dated model\nID in this skill. If an explicit override is needed, verify it is current and\npass the requested model with `-m \u003Cmodel>`.\n\n## If `codex` is not installed\n\nWhen `codex` is missing from PATH, offer to install it — **ask the user for\napproval first**, never install on your own initiative. On yes, follow the\ncurrent instructions at https:\u002F\u002Fdevelopers.openai.com\u002Fcodex\u002Fcli. First-run\nauthentication is interactive — hand that step to the user. Verify with\n`codex --version` before proceeding.\n\n## Prompting Codex\n\nPrompt Codex like an operator, not a collaborator: compact, block-structured\nwith XML tags. State the task, what \"done\" looks like, and the few constraints\nthat matter. A tighter prompt beats a bigger run — improve the contract before\nraising `--effort`.\n\n- **One task per run.** Split unrelated asks (review, then fix, then docs) into\n  separate runs; a mixed prompt gets a mixed result.\n- **Name skills instead of restating them.** Codex reads the same skills your\n  repo carries — say \"follow write-docs for the doc\", \"obey refactor-clean: no\n  compatibility wrappers.\" Don't re-explain what a skill already carries.\n- **Blocks, added only where the task needs them:**\n  - `\u003Ctask>` — the concrete job, the repo\u002Ffailure context, the expected end\n    state. Nearly always present.\n  - `\u003Coutput_contract>` — exact shape, highest-value first, compact.\n  - `\u003Cdefault_follow_through>` — take the low-risk interpretation and keep\n    going; stop only when a missing detail changes correctness, safety, or an\n    irreversible action.\n  - `\u003Cverification_loop>` — before finalizing, check the result against the\n    requirements and the changed files; revise rather than ship the first\n    draft. Any risky fix.\n  - `\u003Cgrounding>` — ground every claim in code or tool output; label inferences\n    as inferences. Review and research.\n  - `\u003Caction_safety>` — keep the diff tightly scoped; no drive-by refactors.\n    Write tasks.\n- **Anti-patterns:** vague framing (\"take a look\"), no output contract (\"report\n  back\"), \"think harder\" in place of a contract, mixing jobs in one run, and\n  demanding certainty the evidence can't support.\n\n## Review — proactive\n\nRun a Codex review whenever you have a substantive diff you'd want a second set\nof eyes on — a refactor, a tricky algorithm, renderer work, a security-sensitive\nchange — before declaring it done or committing. Skip it for trivial edits\n(typos, comments, doc-only).\n\n1. Fix the diff scope before launching: `--uncommitted` for working-tree\n   changes, `--base \u003Cbranch>` for a branch diff, `--commit \u003Csha>` for a landed\n   commit. Resolve a landed commit to its full SHA. Use the non-interactive\n   review entry point with an explicit read-only sandbox and separate progress\n   and verdict files in a fresh temporary directory:\n\n   ```sh\n   review_dir=$(mktemp -d \"${TMPDIR:-\u002Ftmp}\u002Fcodex-review.XXXXXX\")\n   codex exec --sandbox read-only -C \"$(git rev-parse --show-toplevel)\" review \\\n     --commit \"$(git rev-parse HEAD)\" --json -o \"$review_dir\u002Fverdict.md\" \\\n     \u003C \u002Fdev\u002Fnull > \"$review_dir\u002Fevents.jsonl\" 2> \"$review_dir\u002Fstderr.log\"\n   ```\n\n   Background long runs and retain their process\u002Fsession handle. Scope flags\n   and custom instructions are mutually exclusive: a scope flag takes no\n   prompt. For custom framing, specify the exact diff to inspect in the prompt;\n   don't rely on default scope or traverse all refs\u002Fcheckpoint history. Never\n   state the answer you expect (unprimed, same discipline as\n   [screenshot-critique](..\u002F..\u002Fvisual\u002Fscreenshot-critique\u002FSKILL.md)).\n2. Monitor progress in `events.jsonl` and failures in `stderr.log`. An active\n   reviewer reading relevant code is not hung merely because five minutes\n   elapsed. If it stalls or drifts outside the diff and its dependencies,\n   inspect the last command\u002Ferror before interrupting; correct the cause or\n   narrow the task before retrying. Completion requires exit status zero,\n   `turn.completed`, and a nonempty final verdict. A timeout, `turn.failed`, or\n   missing verdict is incomplete, never a clean review.\n3. Triage every finding: confirm it against the code before acting. Preserve\n   Codex's evidence boundaries — an inference it labelled is not a fact.\n   Overlap with your own doubts is high-priority evidence; a finding you\n   dismiss needs a stated reason, not silence.\n4. Report the outcome to the user — what Codex flagged, what you fixed, what\n   you dismissed and why. Done when every finding is either fixed or\n   explicitly dismissed.\n\n## Implementation — explicit ask only\n\nDelegate implementation to Codex only when the user names Codex for the task.\nNever hand it work on your own initiative, and never re-delegate follow-up\nwork without a fresh ask.\n\n1. Slice the task sharp before delegating — goal, constraints, and how to\n   verify — using the prompt discipline above. An underspecified task stays\n   with you until a fresh agent couldn't misread it.\n2. Start from a clean tree (or record the baseline commit) so Codex's diff is\n   separable from yours.\n3. Pick the sandbox by what the task must RUN:\n   - Pure code + typecheck\u002Funit: `codex exec --sandbox workspace-write \"\u003Ctask>\"`.\n     Network is off; add `-c sandbox_workspace_write.network_access=true` only\n     when the task must fetch (e.g. new deps).\n   - **Browser verification, dev servers, or full test runs: use\n     `codex exec --dangerously-bypass-approvals-and-sandbox \"\u003Ctask>\"`.** The\n     sandbox blocks localhost binds (`listen EPERM` on vite\u002Fplaywright), so a\n     sandboxed codex ships code it never saw run. Bypass trades that blindness\n     for zero OS control: only in a dedicated git worktree, only with a prompt\n     you authored end-to-end (never relaying third-party text), and the diff\n     review you owe afterwards is the control.\n   Use `-o \u003Cfile>` to capture the final message and background long calls;\n   suppress codex's stderr thinking-noise with `2>\u002Fdev\u002Fnull` so it doesn't\n   bloat your context (drop it only to debug a failing run), and add\n   `--skip-git-repo-check` to run outside a git repo. Non-interactive runs\n   never ask for approval either way.\n4. Follow up with `codex exec resume \u003Csession-id> \"\u003Cfollow-up>\"`, taking the\n   id from the run header. `resume --last` means the most recent session\n   globally — a review or any other codex run in between will hijack it. If\n   two resume rounds don't converge, stop delegating and finish it yourself —\n   iterating a confused agent costs more than taking over.\n5. You own the result: read the full diff, run the tests, and only then report\n   it. \"Codex says it's done\" is not done.\n\n## Exec liveness — a hang looks like work\n\nA backgrounded `codex exec` can wedge at startup: process alive at ~0% CPU, but\nno session file under `~\u002F.codex\u002Fsessions\u002F\u003CY\u002FM\u002FD>\u002F`, no network socket, no tree\nchanges. \"Process running\" is NOT \"working.\"\n\n- **Stdin is always a file or `\u002Fdev\u002Fnull`, never an inherited terminal.** Left\n  on an interactive\u002Fpiped stdin with no prompt source, exec prints `Reading\n  additional input from stdin...` and blocks forever — the most common hang.\n  Either pass the prompt as a positional arg with `\u003C \u002Fdev\u002Fnull`, or feed a\n  prompt file with `codex exec [flags] - \u003C prompt.txt` (the `-` makes codex\n  read the task from stdin, and a missing file fails the redirect loudly —\n  unlike `\"$(cat prompt.txt)\"`, which silently sends the fallback string as\n  the task). Add `nohup`\u002F`&` as needed.\n- **Watchdog:** put a unique marker in the prompt, then kill the exec if\n  `grep -rl \"\u003Cmarker>\" ~\u002F.codex\u002Fsessions\u002F\u003CY\u002FM\u002FD>\u002F` finds no session within\n  ~3 minutes. Relaunching after a kill reliably works.\n- **Trust the worktree.** Headless exec in a directory Codex doesn't trust can\n  block forever on an invisible prompt. Git worktrees are separate paths from\n  the trusted repo root — add\n  `[projects.\"\u003Cworktree-path>\"]\\ntrust_level = \"trusted\"` to\n  `~\u002F.codex\u002Fconfig.toml` before exec'ing in one. This is the safe fix; the\n  bypass flags stay forbidden here.\n- **Don't launch two execs in the same instant,** and kill stale hung execs\n  before starting a new one.\n- **Launch from the repo\u002Fworktree ROOT.** The writable sandbox root is the\n  CWD at launch: exec'd from a subdirectory (e.g. `web\u002F`), every edit outside\n  it is rejected as \"writing outside of the project\" and a `never` approval\n  policy can't recover — the run burns with zero files changed. Pass\n  `-C \u003Croot>` to set the working root explicitly instead of `cd`-ing into it.\n\n## Rules\n\n- Don't touch the working tree while a Codex exec is running on it.\n- Explicit `--sandbox read-only` for reviews, consultation and questions;\n  `workspace-write` only for delegated implementation.\n- `--dangerously-bypass-approvals-and-sandbox` is reserved for tasks that must\n  run browsers\u002Fservers\u002Ffull suites (above) — dedicated worktree, self-authored\n  prompt, mandatory diff review after. `--full-auto` has been removed (it\n  aliased workspace-write) — don't reach for it.\n- Reasoning effort and model are config\u002Fflag overrides — `-c\n  model_reasoning_effort=high`, `-m \u003Cmodel>` (the old `--effort` flag is gone\n  in current Codex). Leave both at their defaults unless the user asks;\n  tighten the prompt first and keep model IDs unversioned. If a version is\n  required, verify that it is current before using it.\n","# Codex\n\nCodex is an independent agent on PATH (`codex` — invoke as `command codex` if\na shell alias shadows it), sharing this working tree and already authenticated. It is a second opinion, not ground truth: verify what it\nreports, own what it changes. It reads the same skills your repo carries.\n\nLeave Codex's model at its CLI default unless the user asks for a specific\none. The default follows the current Codex release; do not pin a dated model\nID in this skill. If an explicit override is needed, verify it is current and\npass the requested model with `-m \u003Cmodel>`.\n\n## If `codex` is not installed\n\nWhen `codex` is missing from PATH, offer to install it — **ask the user for\napproval first**, never install on your own initiative. On yes, follow the\ncurrent instructions at https:\u002F\u002Fdevelopers.openai.com\u002Fcodex\u002Fcli. First-run\nauthentication is interactive — hand that step to the user. Verify with\n`codex --version` before proceeding.\n\n## Prompting Codex\n\nPrompt Codex like an operator, not a collaborator: compact, block-structured\nwith XML tags. State the task, what \"done\" looks like, and the few constraints\nthat matter. A tighter prompt beats a bigger run — improve the contract before\nraising `--effort`.\n\n- **One task per run.** Split unrelated asks (review, then fix, then docs) into\n  separate runs; a mixed prompt gets a mixed result.\n- **Name skills instead of restating them.** Codex reads the same skills your\n  repo carries — say \"follow write-docs for the doc\", \"obey refactor-clean: no\n  compatibility wrappers.\" Don't re-explain what a skill already carries.\n- **Blocks, added only where the task needs them:**\n  - `\u003Ctask>` — the concrete job, the repo\u002Ffailure context, the expected end\n    state. Nearly always present.\n  - `\u003Coutput_contract>` — exact shape, highest-value first, compact.\n  - `\u003Cdefault_follow_through>` — take the low-risk interpretation and keep\n    going; stop only when a missing detail changes correctness, safety, or an\n    irreversible action.\n  - `\u003Cverification_loop>` — before finalizing, check the result against the\n    requirements and the changed files; revise rather than ship the first\n    draft. Any risky fix.\n  - `\u003Cgrounding>` — ground every claim in code or tool output; label inferences\n    as inferences. Review and research.\n  - `\u003Caction_safety>` — keep the diff tightly scoped; no drive-by refactors.\n    Write tasks.\n- **Anti-patterns:** vague framing (\"take a look\"), no output contract (\"report\n  back\"), \"think harder\" in place of a contract, mixing jobs in one run, and\n  demanding certainty the evidence can't support.\n\n## Review — proactive\n\nRun a Codex review whenever you have a substantive diff you'd want a second set\nof eyes on — a refactor, a tricky algorithm, renderer work, a security-sensitive\nchange — before declaring it done or committing. Skip it for trivial edits\n(typos, comments, doc-only).\n\n1. Fix the diff scope before launching: `--uncommitted` for working-tree\n   changes, `--base \u003Cbranch>` for a branch diff, `--commit \u003Csha>` for a landed\n   commit. Resolve a landed commit to its full SHA. Use the non-interactive\n   review entry point with an explicit read-only sandbox and separate progress\n   and verdict files in a fresh temporary directory:\n\n   ```sh\n   review_dir=$(mktemp -d \"${TMPDIR:-\u002Ftmp}\u002Fcodex-review.XXXXXX\")\n   codex exec --sandbox read-only -C \"$(git rev-parse --show-toplevel)\" review \\\n     --commit \"$(git rev-parse HEAD)\" --json -o \"$review_dir\u002Fverdict.md\" \\\n     \u003C \u002Fdev\u002Fnull > \"$review_dir\u002Fevents.jsonl\" 2> \"$review_dir\u002Fstderr.log\"\n   ```\n\n   Background long runs and retain their process\u002Fsession handle. Scope flags\n   and custom instructions are mutually exclusive: a scope flag takes no\n   prompt. For custom framing, specify the exact diff to inspect in the prompt;\n   don't rely on default scope or traverse all refs\u002Fcheckpoint history. Never\n   state the answer you expect (unprimed, same discipline as\n   [screenshot-critique](..\u002F..\u002Fvisual\u002Fscreenshot-critique\u002FSKILL.md)).\n2. Monitor progress in `events.jsonl` and failures in `stderr.log`. An active\n   reviewer reading relevant code is not hung merely because five minutes\n   elapsed. If it stalls or drifts outside the diff and its dependencies,\n   inspect the last command\u002Ferror before interrupting; correct the cause or\n   narrow the task before retrying. Completion requires exit status zero,\n   `turn.completed`, and a nonempty final verdict. A timeout, `turn.failed`, or\n   missing verdict is incomplete, never a clean review.\n3. Triage every finding: confirm it against the code before acting. Preserve\n   Codex's evidence boundaries — an inference it labelled is not a fact.\n   Overlap with your own doubts is high-priority evidence; a finding you\n   dismiss needs a stated reason, not silence.\n4. Report the outcome to the user — what Codex flagged, what you fixed, what\n   you dismissed and why. Done when every finding is either fixed or\n   explicitly dismissed.\n\n## Implementation — explicit ask only\n\nDelegate implementation to Codex only when the user names Codex for the task.\nNever hand it work on your own initiative, and never re-delegate follow-up\nwork without a fresh ask.\n\n1. Slice the task sharp before delegating — goal, constraints, and how to\n   verify — using the prompt discipline above. An underspecified task stays\n   with you until a fresh agent couldn't misread it.\n2. Start from a clean tree (or record the baseline commit) so Codex's diff is\n   separable from yours.\n3. Pick the sandbox by what the task must RUN:\n   - Pure code + typecheck\u002Funit: `codex exec --sandbox workspace-write \"\u003Ctask>\"`.\n     Network is off; add `-c sandbox_workspace_write.network_access=true` only\n     when the task must fetch (e.g. new deps).\n   - **Browser verification, dev servers, or full test runs: use\n     `codex exec --dangerously-bypass-approvals-and-sandbox \"\u003Ctask>\"`.** The\n     sandbox blocks localhost binds (`listen EPERM` on vite\u002Fplaywright), so a\n     sandboxed codex ships code it never saw run. Bypass trades that blindness\n     for zero OS control: only in a dedicated git worktree, only with a prompt\n     you authored end-to-end (never relaying third-party text), and the diff\n     review you owe afterwards is the control.\n   Use `-o \u003Cfile>` to capture the final message and background long calls;\n   suppress codex's stderr thinking-noise with `2>\u002Fdev\u002Fnull` so it doesn't\n   bloat your context (drop it only to debug a failing run), and add\n   `--skip-git-repo-check` to run outside a git repo. Non-interactive runs\n   never ask for approval either way.\n4. Follow up with `codex exec resume \u003Csession-id> \"\u003Cfollow-up>\"`, taking the\n   id from the run header. `resume --last` means the most recent session\n   globally — a review or any other codex run in between will hijack it. If\n   two resume rounds don't converge, stop delegating and finish it yourself —\n   iterating a confused agent costs more than taking over.\n5. You own the result: read the full diff, run the tests, and only then report\n   it. \"Codex says it's done\" is not done.\n\n## Exec liveness — a hang looks like work\n\nA backgrounded `codex exec` can wedge at startup: process alive at ~0% CPU, but\nno session file under `~\u002F.codex\u002Fsessions\u002F\u003CY\u002FM\u002FD>\u002F`, no network socket, no tree\nchanges. \"Process running\" is NOT \"working.\"\n\n- **Stdin is always a file or `\u002Fdev\u002Fnull`, never an inherited terminal.** Left\n  on an interactive\u002Fpiped stdin with no prompt source, exec prints `Reading\n  additional input from stdin...` and blocks forever — the most common hang.\n  Either pass the prompt as a positional arg with `\u003C \u002Fdev\u002Fnull`, or feed a\n  prompt file with `codex exec [flags] - \u003C prompt.txt` (the `-` makes codex\n  read the task from stdin, and a missing file fails the redirect loudly —\n  unlike `\"$(cat prompt.txt)\"`, which silently sends the fallback string as\n  the task). Add `nohup`\u002F`&` as needed.\n- **Watchdog:** put a unique marker in the prompt, then kill the exec if\n  `grep -rl \"\u003Cmarker>\" ~\u002F.codex\u002Fsessions\u002F\u003CY\u002FM\u002FD>\u002F` finds no session within\n  ~3 minutes. Relaunching after a kill reliably works.\n- **Trust the worktree.** Headless exec in a directory Codex doesn't trust can\n  block forever on an invisible prompt. Git worktrees are separate paths from\n  the trusted repo root — add\n  `[projects.\"\u003Cworktree-path>\"]\\ntrust_level = \"trusted\"` to\n  `~\u002F.codex\u002Fconfig.toml` before exec'ing in one. This is the safe fix; the\n  bypass flags stay forbidden here.\n- **Don't launch two execs in the same instant,** and kill stale hung execs\n  before starting a new one.\n- **Launch from the repo\u002Fworktree ROOT.** The writable sandbox root is the\n  CWD at launch: exec'd from a subdirectory (e.g. `web\u002F`), every edit outside\n  it is rejected as \"writing outside of the project\" and a `never` approval\n  policy can't recover — the run burns with zero files changed. Pass\n  `-C \u003Croot>` to set the working root explicitly instead of `cd`-ing into it.\n\n## Rules\n\n- Don't touch the working tree while a Codex exec is running on it.\n- Explicit `--sandbox read-only` for reviews, consultation and questions;\n  `workspace-write` only for delegated implementation.\n- `--dangerously-bypass-approvals-and-sandbox` is reserved for tasks that must\n  run browsers\u002Fservers\u002Ffull suites (above) — dedicated worktree, self-authored\n  prompt, mandatory diff review after. `--full-auto` has been removed (it\n  aliased workspace-write) — don't reach for it.\n- Reasoning effort and model are config\u002Fflag overrides — `-c\n  model_reasoning_effort=high`, `-m \u003Cmodel>` (the old `--effort` flag is gone\n  in current Codex). Leave both at their defaults unless the user asks;\n  tighten the prompt first and keep model IDs unversioned. If a version is\n  required, verify that it is current before using it.\n",{"name":5,"description":7},"main","fc9ff73613f4346442defd1318eae264af415d72","provenance\u002Fupstream\u002Fcodex\u002Fskill",1,"https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit","https:\u002F\u002Ftransitivebullsh.it","My heart is open source 💕 ","2026-10-24 01:38:48",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add transitive-bullshit\u002Fskills --skill codex",{"tool":39,"command":40},"bunx","bunx skills add transitive-bullshit\u002Fskills --skill codex",{"tool":42,"command":43},"pnpm","pnpm dlx skills add transitive-bullshit\u002Fskills --skill codex","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fprovenance\u002Fupstream\u002Fcodex\u002Fskill\u002FSKILL.md",true,[],[48,55,62,70,77,84],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M7C8Z0JQ0VH6YYSJQGXJKVV4","xurl","Paid X API fallback through xurl after cheaper X reads fail, or for authorized account actions unsupported by bird.","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fxurl\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":59,"categories":60,"keywords":61,"publishedAt":19},"01M7C8YY74K2MJYK311R1PG31A","x-data","Read or research X\u002FTwitter posts, URLs, profiles, threads, or the user's X history using the cheapest available source.","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fx-data\u002FSKILL.md",[],[],{"skillNo":63,"name":64,"nameZh":6,"description":65,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":66,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7C8YV6AAFSCTS39C2MBKTTV","write-skills","Create or revise agent skills. Use when adding a new skill file, renaming a skill, simplifying an existing skill, improving trigger descriptions, or deciding what belongs in a skill versus references, scripts, assets, or ordinary docs.","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwrite-skills\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwrite-skills\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7C8YQFGTFVQQJCH8SPFS3YQ","web-perf","Audit, diagnose, or optimize website loading and interaction performance, Core Web Vitals, and Lighthouse performance scores.","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fweb-perf\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7C8YJ6QFV6QFCJESHGZZ2B8","vercel-react-best-practices","Apply React and Next.js performance guidance. Use for performance work or a specific rendering, data-fetching, or bundle concern.","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fvercel-react-best-practices\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7C8YBRVCMAZ99D1Y5QXHMG8","vercel-optimize","Audit deployed Vercel cost and performance using project metrics. Use for expensive or slow routes, usage analysis, and caching opportunities.","https:\u002F\u002Fgithub.com\u002Ftransitive-bullshit\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fvercel-optimize\u002FSKILL.md",[],[]]