[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75VT76B7CZ668ZGD802WWVK":3,"skill-same:01M75VT76B7CZ668ZGD802WWVK":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},"01M75VT76B7CZ668ZGD802WWVK","pr-monitor",null,"Handle one pull request attention event end to end — read the watcher digest, triage it, delegate the fixing to pr-fix, re-observe the pull request, and exit. Use when the PR watcher wakes you, or run it manually for a one-shot check of an open PR.","ronaknnathani\u002Frelay","ronaknnathani","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F7279934?v=4","01M75VS2S5V1DVP31PCWYA7SMA","Ronak Nathani","01M75VS2STNTX5Y5JGYQGRE0NG",44,"MIT","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Fpr-monitor\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: pr-monitor\ndescription: Handle one pull request attention event end to end — read the watcher digest, triage it, delegate the fixing to pr-fix, re-observe the pull request, and exit. Use when the PR watcher wakes you, or run it manually for a one-shot check of an open PR.\n---\n\n# PR Monitor\n\nHandle **one** pull request attention event, then stop. You are a **router**: the `relay pr watch`\nruntime does the observing, you interpret its digest, delegate the fixing to `pr-fix`, re-observe the\npull request, and exit. You never write code, never post in your own voice, never approve, and never\nmerge.\n\nThere is **no acknowledgement**. The watcher never records that attention was handled, because a local\nclaim can be wrong. Every tick re-reads the live pull request, and an item disappears only when the\nremote state itself no longer shows it: the check passes or reruns, the head SHA changes, the conflict\nclears, the thread resolves, your reply lands on that exact source, auto-merge is armed, or the pull\nrequest closes or merges. If nothing changed, the problem is still there and it will wake you again —\nwhich is correct.\n\nThis skill has **no loop**. It does not schedule, does not recur, and does not own a next-tick time.\nThe watcher owns cadence; one run of this skill owns one digest.\n\n## Where the work comes from\n\n**Woken by the watcher.** The prompt names the project and a fingerprint:\n`Run pr-monitor once for project \u003Cslug> using watcher fingerprint \u003Cfp>.` Bind them to `$SLUG` and\n`$FP`, then read the digest that fingerprint names:\n\n```bash\nrelay pr watch digest \"$SLUG\" --fingerprint \"$FP\" --json\nrelay pr watch status \"$SLUG\" --json     # current fingerprint, cadence, last wake, suppression\n```\n\nThe digest record is refreshed on every observation, so the pull request metadata and bodies it\ncarries are the newest the watcher saw for that fingerprint — never a stale snapshot.\n\n**Invoked manually, or with no Herdr.** Observe first, then act on what that observation recorded:\n\n```bash\nrelay pr watch tick \"$SLUG\" --json       # fresh read-only observation; records its digest\n```\n\n`tick` works with no watcher running and never changes the watcher's schedule. If it returns no\nactionable items, there is nothing to do — say so and stop.\n\n## What the digest already decided\n\nThe digest is deterministic and complete; do not re-derive it with your own `gh` sweep. It carries the\npull request state, draft flag, base and head refs and SHAs, `mergeStateStatus`, `mergeable`,\n`reviewDecision`, whether auto-merge is armed, every check with its status, conclusion, and run id, and\nevery actionable item with its `source`, `id`, `answers` token, `updatedAt`, `body`, thread id, file,\nand line.\n\nEach item carries a `reason`:\n\n| Reason | What it means | Who acts |\n|---|---|---|\n| `failing-check` | a check concluded failure, error, timeout, action-required, startup-failure, canceled, or stale | you classify flake vs real; real → `pr-fix`. A canceled or stale check reported no result at all — rerun it |\n| `changes-requested` | a CHANGES_REQUESTED review nobody has answered yet | `pr-fix` — answer that exact review; once answered the watcher reports `changes-requested-awaiting-rereview` and waits for the reviewer |\n| `new-comment` \u002F `new-review` \u002F `new-inline-comment` | human activity the agent has not answered | `pr-fix` |\n| `unresolved-thread` | an unresolved thread still holding human activity no reply of yours names | `pr-fix` — a resolved thread is never reported, so resolving one is how it clears |\n| `merge-conflict` | the branch conflicts with its base | `pr-fix` |\n| `blocked` | GitHub will not merge it and nothing else in the digest explains why | you investigate branch protection and required checks, then escalate |\n| `stale-base` | the branch is behind its base | you rebase |\n| `auto-merge-not-armed` | approved, green, clean, based on the default branch, not armed | you arm it |\n| `closed-unmerged` | the pull request was closed without merging | escalate to the author |\n| `stack-front-merged` | the stack's front pull request merged | the stack orchestrator retargets |\n\nFetch more only for the specific item you are acting on — a failed run's log, a file's history. Never\nre-page the whole comment history; the watcher already did, and its digest is the record.\n\n## The run\n\n### 1. Triage\n\nSplit the items into what you own and what `pr-fix` owns.\n\n**You classify each `failing-check` as an infra flake or a real failure.** An infra flake is a\ndependency download, TLS timeout, registry 5xx, or sandbox limit — inspect the failed run\n(`gh run view \u003Crun-id> --log-failed`, using the run id the digest carries) before deciding. Everything\nelse is real.\n\n### 2. Delegate the real work to `pr-fix` — one writer, one call\n\nHand `pr-fix` **one** scoped worklist as a sub-agent: the real check failures with their names and run\nids, the comment, review, and thread items **with the bodies, ids, `answers` tokens, thread ids, files,\nand lines the digest already carries**, and any merge conflict. Tell it that it is running in\n**delegated mode**: the worklist is complete, it must not re-fetch the broad pull request context and\nmust not loop, and it must return a structured per-item result.\n\nPass each item's `answers` token through verbatim and require every reply to carry it in its marker.\nThat token is how the watcher knows which single piece of feedback was answered; without it, a reply\neither answers nothing or — worse, if the watcher went back to matching on time — buries a comment\nwritten after the observation you are acting on and shown to nobody.\n\nNever run two writers on one branch: exactly one `pr-fix` call per run, and never hand it an infra\nflake.\n\n### 3. Do the agent-side actions the digest names\n\nThese change no code, so they are yours — run them after `pr-fix` returns, so pushes stay serialized:\n\n- **a `failing-check` you classified as a flake** → `gh run rerun \u003Crun-id> --failed` (never cancel a\n  queued run).\n- **`stale-base`** → rebase onto the fresh base so a new head re-triggers the checks, then force-push\n  with `--force-with-lease`. If the rebase surfaces a conflict it is no longer staleness — leave it for\n  `pr-fix` on the next attention event.\n- **`auto-merge-not-armed`** → re-arm auto-merge. It silently turns off after a force-push and after a\n  CHANGES_REQUESTED→APPROVED transition. Arm it **only** on a pull request based on the default branch;\n  \"already queued\" means it is armed.\n- **`closed-unmerged`** → do not reopen it; surface it to the author as a durable escalation.\n- **`stack-front-merged`** → report it to the stack orchestrator; the front-advance is its job.\n\n### 4. Re-observe once — the remote state is the record\n\n```bash\nrelay pr watch tick \"$SLUG\" --json\n```\n\nThis is the only \"did it work\" signal there is. Compare it to the digest you started from:\n\n- an item that is **gone** was genuinely resolved on GitHub;\n- an item that is **still there** was not, whatever `pr-fix` reported.\n\nIf `pr-fix` failed or came back partial, say exactly what remains. There is nothing to suppress and\nnothing to record: the next scheduled check re-observes and brings back whatever is still true.\n\n### 5. Report, exit\n\nReport a one-screen digest: what the attention was, what was delegated, what you did yourself, and\nwhat the re-observation still shows. Then **stop**. Do not wait, do not schedule, do not start another\ncycle.\n\nIf the pull request was **closed without merging**, the watcher hands you that escalation once and\nthen finishes — surface it to the author. If a merged **stack front** woke you, the stack orchestrator\nowns the front-advance and must stop the old watcher with `relay pr watch stop \u003Cfront-project-slug>`.\n\n## Under a project workflow\n\nRecord the outcome with `relay state log \u003Cslug> \"\u003Cone-line digest>\"` when the project has Relay state.\nIf no watcher is running and this project should be watched, start it once with\n`relay pr watch start \u003Cslug>` from a Herdr pane. Never start a second watcher for a project — `start`\nadopts the running one. It refuses outright, creating nothing, unless exactly one live session carries\nthat project's Relay identity; if it refuses, say so and keep using this skill by hand.\n\n## Guardrails (non-negotiable)\n\n- **No loop.** One digest, one run, one exit. Never claim continuous monitoring.\n- **No local \"handled\" claim.** The remote pull request is the only record of what is resolved.\n- **Approval is the only merge path.** Never self-approve, never `gh pr merge` to merge now, never\n  dismiss a review to unblock. Auto-merge fires on a genuine human code-owner approval.\n- **Never impersonate, and always mark.** Every automated reply — yours or `pr-fix`'s — opens with the\n  exact hidden marker `\u003C!-- relay-agent-reply answers=\u003Citem answers token> -->` on its own\n  line, then the visible `🤖 \u003Cagent> on behalf of \u003Cauthor>` disclosure. The marker is the only thing\n  that tells the watcher an agent replied, and the token is the only thing that tells it *what* was\n  answered; a reply without either is read as new human feedback forever. Reply on the **same source**\n  you are answering: a conversation comment with `gh pr comment`, a review body with\n  `gh pr review --comment`, an inline comment or thread with the inline replies endpoint.\n- **Never silence a failure** (enforced inside `pr-fix`).\n- **One writer per branch** — serialize every push.\n- **Never report an item as resolved** unless the re-observation stopped showing it.\n\n## Red flags\n\n- Running a broad `gh` comment\u002Fcheck sweep instead of reading the digest the watcher already recorded.\n- Reporting work as done on `pr-fix`'s word instead of on the re-observation.\n- Handing `pr-fix` an infra flake, or rerunning a check it is already fixing.\n- Two sub-agents pushing the same branch in one run.\n- Letting a reply go out without the item's exact `answers` token, or with an id you made up.\n- Scheduling a follow-up tick, recording a next-tick time, or starting a loop.\n- Arming auto-merge on a pull request that is not based on the default branch.\n\n## Verification checklist\n\n- [ ] Read exactly one digest — by fingerprint from the wake, or from a fresh `tick`.\n- [ ] Every `failing-check` was classified flake vs real before anything was delegated or rerun.\n- [ ] Real failures, comments, threads, and conflicts went to a single delegated `pr-fix` call carrying their bodies, ids, and `answers` tokens.\n- [ ] Flake reruns, the stale rebase, and auto-merge re-arming happened here, after `pr-fix` returned.\n- [ ] One `relay pr watch tick` ran and every claim of \"resolved\" is backed by an item it no longer reports.\n- [ ] The outcome was reported and the run exited without scheduling anything.\n","# PR Monitor\n\nHandle **one** pull request attention event, then stop. You are a **router**: the `relay pr watch`\nruntime does the observing, you interpret its digest, delegate the fixing to `pr-fix`, re-observe the\npull request, and exit. You never write code, never post in your own voice, never approve, and never\nmerge.\n\nThere is **no acknowledgement**. The watcher never records that attention was handled, because a local\nclaim can be wrong. Every tick re-reads the live pull request, and an item disappears only when the\nremote state itself no longer shows it: the check passes or reruns, the head SHA changes, the conflict\nclears, the thread resolves, your reply lands on that exact source, auto-merge is armed, or the pull\nrequest closes or merges. If nothing changed, the problem is still there and it will wake you again —\nwhich is correct.\n\nThis skill has **no loop**. It does not schedule, does not recur, and does not own a next-tick time.\nThe watcher owns cadence; one run of this skill owns one digest.\n\n## Where the work comes from\n\n**Woken by the watcher.** The prompt names the project and a fingerprint:\n`Run pr-monitor once for project \u003Cslug> using watcher fingerprint \u003Cfp>.` Bind them to `$SLUG` and\n`$FP`, then read the digest that fingerprint names:\n\n```bash\nrelay pr watch digest \"$SLUG\" --fingerprint \"$FP\" --json\nrelay pr watch status \"$SLUG\" --json     # current fingerprint, cadence, last wake, suppression\n```\n\nThe digest record is refreshed on every observation, so the pull request metadata and bodies it\ncarries are the newest the watcher saw for that fingerprint — never a stale snapshot.\n\n**Invoked manually, or with no Herdr.** Observe first, then act on what that observation recorded:\n\n```bash\nrelay pr watch tick \"$SLUG\" --json       # fresh read-only observation; records its digest\n```\n\n`tick` works with no watcher running and never changes the watcher's schedule. If it returns no\nactionable items, there is nothing to do — say so and stop.\n\n## What the digest already decided\n\nThe digest is deterministic and complete; do not re-derive it with your own `gh` sweep. It carries the\npull request state, draft flag, base and head refs and SHAs, `mergeStateStatus`, `mergeable`,\n`reviewDecision`, whether auto-merge is armed, every check with its status, conclusion, and run id, and\nevery actionable item with its `source`, `id`, `answers` token, `updatedAt`, `body`, thread id, file,\nand line.\n\nEach item carries a `reason`:\n\n| Reason | What it means | Who acts |\n|---|---|---|\n| `failing-check` | a check concluded failure, error, timeout, action-required, startup-failure, canceled, or stale | you classify flake vs real; real → `pr-fix`. A canceled or stale check reported no result at all — rerun it |\n| `changes-requested` | a CHANGES_REQUESTED review nobody has answered yet | `pr-fix` — answer that exact review; once answered the watcher reports `changes-requested-awaiting-rereview` and waits for the reviewer |\n| `new-comment` \u002F `new-review` \u002F `new-inline-comment` | human activity the agent has not answered | `pr-fix` |\n| `unresolved-thread` | an unresolved thread still holding human activity no reply of yours names | `pr-fix` — a resolved thread is never reported, so resolving one is how it clears |\n| `merge-conflict` | the branch conflicts with its base | `pr-fix` |\n| `blocked` | GitHub will not merge it and nothing else in the digest explains why | you investigate branch protection and required checks, then escalate |\n| `stale-base` | the branch is behind its base | you rebase |\n| `auto-merge-not-armed` | approved, green, clean, based on the default branch, not armed | you arm it |\n| `closed-unmerged` | the pull request was closed without merging | escalate to the author |\n| `stack-front-merged` | the stack's front pull request merged | the stack orchestrator retargets |\n\nFetch more only for the specific item you are acting on — a failed run's log, a file's history. Never\nre-page the whole comment history; the watcher already did, and its digest is the record.\n\n## The run\n\n### 1. Triage\n\nSplit the items into what you own and what `pr-fix` owns.\n\n**You classify each `failing-check` as an infra flake or a real failure.** An infra flake is a\ndependency download, TLS timeout, registry 5xx, or sandbox limit — inspect the failed run\n(`gh run view \u003Crun-id> --log-failed`, using the run id the digest carries) before deciding. Everything\nelse is real.\n\n### 2. Delegate the real work to `pr-fix` — one writer, one call\n\nHand `pr-fix` **one** scoped worklist as a sub-agent: the real check failures with their names and run\nids, the comment, review, and thread items **with the bodies, ids, `answers` tokens, thread ids, files,\nand lines the digest already carries**, and any merge conflict. Tell it that it is running in\n**delegated mode**: the worklist is complete, it must not re-fetch the broad pull request context and\nmust not loop, and it must return a structured per-item result.\n\nPass each item's `answers` token through verbatim and require every reply to carry it in its marker.\nThat token is how the watcher knows which single piece of feedback was answered; without it, a reply\neither answers nothing or — worse, if the watcher went back to matching on time — buries a comment\nwritten after the observation you are acting on and shown to nobody.\n\nNever run two writers on one branch: exactly one `pr-fix` call per run, and never hand it an infra\nflake.\n\n### 3. Do the agent-side actions the digest names\n\nThese change no code, so they are yours — run them after `pr-fix` returns, so pushes stay serialized:\n\n- **a `failing-check` you classified as a flake** → `gh run rerun \u003Crun-id> --failed` (never cancel a\n  queued run).\n- **`stale-base`** → rebase onto the fresh base so a new head re-triggers the checks, then force-push\n  with `--force-with-lease`. If the rebase surfaces a conflict it is no longer staleness — leave it for\n  `pr-fix` on the next attention event.\n- **`auto-merge-not-armed`** → re-arm auto-merge. It silently turns off after a force-push and after a\n  CHANGES_REQUESTED→APPROVED transition. Arm it **only** on a pull request based on the default branch;\n  \"already queued\" means it is armed.\n- **`closed-unmerged`** → do not reopen it; surface it to the author as a durable escalation.\n- **`stack-front-merged`** → report it to the stack orchestrator; the front-advance is its job.\n\n### 4. Re-observe once — the remote state is the record\n\n```bash\nrelay pr watch tick \"$SLUG\" --json\n```\n\nThis is the only \"did it work\" signal there is. Compare it to the digest you started from:\n\n- an item that is **gone** was genuinely resolved on GitHub;\n- an item that is **still there** was not, whatever `pr-fix` reported.\n\nIf `pr-fix` failed or came back partial, say exactly what remains. There is nothing to suppress and\nnothing to record: the next scheduled check re-observes and brings back whatever is still true.\n\n### 5. Report, exit\n\nReport a one-screen digest: what the attention was, what was delegated, what you did yourself, and\nwhat the re-observation still shows. Then **stop**. Do not wait, do not schedule, do not start another\ncycle.\n\nIf the pull request was **closed without merging**, the watcher hands you that escalation once and\nthen finishes — surface it to the author. If a merged **stack front** woke you, the stack orchestrator\nowns the front-advance and must stop the old watcher with `relay pr watch stop \u003Cfront-project-slug>`.\n\n## Under a project workflow\n\nRecord the outcome with `relay state log \u003Cslug> \"\u003Cone-line digest>\"` when the project has Relay state.\nIf no watcher is running and this project should be watched, start it once with\n`relay pr watch start \u003Cslug>` from a Herdr pane. Never start a second watcher for a project — `start`\nadopts the running one. It refuses outright, creating nothing, unless exactly one live session carries\nthat project's Relay identity; if it refuses, say so and keep using this skill by hand.\n\n## Guardrails (non-negotiable)\n\n- **No loop.** One digest, one run, one exit. Never claim continuous monitoring.\n- **No local \"handled\" claim.** The remote pull request is the only record of what is resolved.\n- **Approval is the only merge path.** Never self-approve, never `gh pr merge` to merge now, never\n  dismiss a review to unblock. Auto-merge fires on a genuine human code-owner approval.\n- **Never impersonate, and always mark.** Every automated reply — yours or `pr-fix`'s — opens with the\n  exact hidden marker `\u003C!-- relay-agent-reply answers=\u003Citem answers token> -->` on its own\n  line, then the visible `🤖 \u003Cagent> on behalf of \u003Cauthor>` disclosure. The marker is the only thing\n  that tells the watcher an agent replied, and the token is the only thing that tells it *what* was\n  answered; a reply without either is read as new human feedback forever. Reply on the **same source**\n  you are answering: a conversation comment with `gh pr comment`, a review body with\n  `gh pr review --comment`, an inline comment or thread with the inline replies endpoint.\n- **Never silence a failure** (enforced inside `pr-fix`).\n- **One writer per branch** — serialize every push.\n- **Never report an item as resolved** unless the re-observation stopped showing it.\n\n## Red flags\n\n- Running a broad `gh` comment\u002Fcheck sweep instead of reading the digest the watcher already recorded.\n- Reporting work as done on `pr-fix`'s word instead of on the re-observation.\n- Handing `pr-fix` an infra flake, or rerunning a check it is already fixing.\n- Two sub-agents pushing the same branch in one run.\n- Letting a reply go out without the item's exact `answers` token, or with an id you made up.\n- Scheduling a follow-up tick, recording a next-tick time, or starting a loop.\n- Arming auto-merge on a pull request that is not based on the default branch.\n\n## Verification checklist\n\n- [ ] Read exactly one digest — by fingerprint from the wake, or from a fresh `tick`.\n- [ ] Every `failing-check` was classified flake vs real before anything was delegated or rerun.\n- [ ] Real failures, comments, threads, and conflicts went to a single delegated `pr-fix` call carrying their bodies, ids, and `answers` tokens.\n- [ ] Flake reruns, the stale rebase, and auto-merge re-arming happened here, after `pr-fix` returned.\n- [ ] One `relay pr watch tick` ran and every claim of \"resolved\" is backed by an item it no longer reports.\n- [ ] The outcome was reported and the run exited without scheduling anything.\n",{"name":5,"description":7},"main","5ab51c2d8af79f04294f91621a5b65e21f39dd53","skills\u002Fpr-monitor",4,"https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay","https:\u002F\u002Fgithub.com\u002Fronaknnathani","ronaknathani.com","Compute Infrastructure | Kubernetes @ LinkedIn","2026-11-06 14:25:14",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add ronaknnathani\u002Frelay --skill pr-monitor",{"tool":39,"command":40},"bunx","bunx skills add ronaknnathani\u002Frelay --skill pr-monitor",{"tool":42,"command":43},"pnpm","pnpm dlx skills add ronaknnathani\u002Frelay --skill pr-monitor","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Fpr-monitor\u002FSKILL.md",true,[],[48,55,62,69,76,83],{"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},"01M75VVBP3R7RHF93R3RFWC3SE","write-like-me","Write or edit prose in author's voice, or check that a draft sounds like them. Use for blog posts, design and strategy docs, PR descriptions, commits and review comments, chat \u002F agent instructions, peer feedback, or any first-person writing on their behalf, and whenever an agent is producing text that will go out under their name and needs to sound like them, not like a generated reply.","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Fwrite-like-me\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},"01M75VV70MDT9DZWQG1ZCA3VH6","validate","Verify a change is ready to ship by checking it against its acceptance criteria and running the repo's own quality gates in order, then report a pass\u002Ffail verdict with the exact failures. Use to decide go\u002Fno-go right before shipping — when you want a pass\u002Ffail verdict against acceptance criteria and the repo's quality gates (lint\u002Fbuild\u002Ftest), not a code critique (use `review` for that). It reports — it does not fix (route failures to `pr-fix`\u002F`implement`) and does not ship.","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Fvalidate\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":6,"sourceHtmlUrl":66,"categories":67,"keywords":68,"publishedAt":19},"01M75VV5DEBH4ZS4XDY27JGHCF","todo","Capture, list, and complete repo-scoped todos","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Ftodo\u002FSKILL.md",[],[],{"skillNo":70,"name":71,"nameZh":6,"description":72,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":73,"categories":74,"keywords":75,"publishedAt":19},"01M75VV38NYR4V51C27WPMFG5F","tl","Manage a Relay program as the CEO-facing tech lead: shape goals and contracts, prioritize dependency-aware work, dispatch senior worker agents, and surface every decision without writing production code or merging.","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Ftl\u002FSKILL.md",[],[],{"skillNo":77,"name":78,"nameZh":6,"description":79,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M75VTY2ZTZYJ09AYP6N46FRT","stack-ship","Use only for a standalone stack explicitly launched with `relay --workflow stack-ship`; never invoke inside a tech-lead-managed Relay program. Decompose one goal into small PRs, delegate build and monitor work, surface author decisions, and auto-merge only after human approval.","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Fstack-ship\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":86,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":87,"categories":88,"keywords":89,"publishedAt":19},"01M75VTS6YN6082J93ASWJKENW","simplify","Reduce complexity in recently changed code — cut needless abstraction, dead code, and duplication; improve names and structure — WITHOUT changing behavior. Quality only; it does not hunt for bugs (use review for that). Use after implementing a chunk of work, before review, or whenever a diff feels more complicated than the problem it solves. Triggers on \"clean this up\" \u002F \"tidy this\" \u002F \"refactor this\" when code already exists; for an underspecified goal with no code yet, use `clarify`.","https:\u002F\u002Fgithub.com\u002Fronaknnathani\u002Frelay\u002Fblob\u002Fmain\u002Fskills\u002Fsimplify\u002FSKILL.md",[],[]]