[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M74FSG410A2H12VGX8D94QJ6":3,"skill-same:01M74FSG410A2H12VGX8D94QJ6":77},{"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":38,"commitSha":39,"skillDirectory":40,"forksCount":41,"repositoryHtmlUrl":42,"repositoryHomepageUrl":6,"ownerHtmlUrl":43,"ownerBlogUrl":44,"ownerDescription":45,"updatedAt":46,"install":47,"resources":60},"01M74FSG410A2H12VGX8D94QJ6","handoff",null,"Hands in-progress work from one agent or session to another through a structured handoff document committed to the repo under `.agents\u002Fhandoffs\u002F`, and resumes an existing handoff by its number. Captures goal, context gathered, progress, next steps and open questions in a document that assumes none of the writer's session, so the work can continue in a fresh session or on another machine. The resuming agent deletes the handoff once the work is finished — not when it reads it. Use when the user wants to hand off, park or pause work for another agent or session, to write or read a handoff, to continue handed-off work, or says things like \"hand this off\", \"write a handoff\", \"continue handoff 0003\", \"resume the handoff\", \"übergib das\", \"fahre das fort\".","TitusKirch\u002Fskills","TitusKirch","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F16943133?v=4","01M74FN1AAM5K2NH5EG292B2HZ","Titus Kirch","01M74FN1AX7TX6RAECBHE5PMNT",3,"MIT","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fhandoff\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: handoff\nmetadata:\n  summary: Hands off in-progress work to another agent or session via a committed handoff document.\ndescription: Hands in-progress work from one agent or session to another through a structured handoff document committed to the repo under `.agents\u002Fhandoffs\u002F`, and resumes an existing handoff by its number. Captures goal, context gathered, progress, next steps and open questions in a document that assumes none of the writer's session, so the work can continue in a fresh session or on another machine. The resuming agent deletes the handoff once the work is finished — not when it reads it. Use when the user wants to hand off, park or pause work for another agent or session, to write or read a handoff, to continue handed-off work, or says things like \"hand this off\", \"write a handoff\", \"continue handoff 0003\", \"resume the handoff\", \"übergib das\", \"fahre das fort\".\nallowed-tools:\n  - Read\n  - Write\n  - Edit\n  - Grep\n  - Glob\n  - Bash(jq:*)\n  - Bash(printf:*)\n  - Bash(git log:*)\n  - Bash(git rev-parse:*)\n  - Bash(ls:*)\n  - Bash(tail:*)\n---\n\n# handoff\n\nMove **in-progress work** from one agent to the next — a different session, a different machine, a different day — through a **structured document committed to the repo**. Two directions, one skill: **write** a handoff, or **resume** one.\n\nThe document is the whole product. It is written for a reader who has **none of the writer's session** — no scrollback, no open files, no memory of what was already tried. A handoff that only makes sense to the agent that wrote it has failed at the one job it has.\n\n## Workflow\n\n### 1. Route\n\n| Situation                                                             | Action                       |\n| :-------------------------------------------------------------------- | :--------------------------- |\n| Work is in flight and must be parked or passed on                     | **Write** — steps 2–5        |\n| A handoff exists and the user wants to continue it (`0003`, \"resume\") | **Resume** — steps 6–8       |\n| Work was resumed but is unfinished and is being parked again          | **Update in place** — step 9 |\n\n### 2. Make the work reachable — before writing anything\n\nA handoff is committed **so another machine can pick it up**. That promise is only kept if the _work_ travels too, not just the note about it.\n\n- **Uncommitted work is unreachable.** Commit it (via `atomic-commit`) and push the branch. Work in progress that is not yet coherent goes in one honest `wip` commit — the resumer needs the code more than it needs a tidy history.\n- **Nothing to commit** is a fine answer — say so in the document (`Progress` records \"nothing uncommitted\").\n- **Cannot push** (no remote, offline, work that must not leave the machine) → still write the handoff, but state it plainly under `Progress` and drop the `branch` field. A handoff describing work that exists nowhere but this disk is a **local note**, and must not claim otherwise.\n\n### 3. Allocate the id\n\nHandoffs live in **`.agents\u002Fhandoffs\u002F`**, named `NNNN-lower-kebab-case.md` (`0001-refactor-auth-layer.md`).\n\n**The next id is one above the highest ever used — read from git history, not the folder.** Handoffs are deleted when their work completes, so the folder is a poor witness; history is the only one that remembers `0003` existed. Ids are never reused, renumbered or gap-filled. Recipe: [REFERENCE.md](REFERENCE.md#id-allocation).\n\n### 4. Write the document\n\nCopy [`templates\u002Fhandoff.md`](templates\u002Fhandoff.md) and fill every section. Frontmatter carries the machine-readable fields (`title`, `status`, `created`, `updated`, `branch`, `issue`); the body carries the prose, in fixed H2s: `Goal`, `Context`, `Progress`, `Next steps`, `Open questions`. Full contract: [REFERENCE.md](REFERENCE.md#document-contract).\n\nWrite it **self-contained**:\n\n- **No deixis.** No \"the file we looked at\", \"as discussed\", \"the failing test from earlier\". Name paths, symbols, commands, issue numbers and commit shas outright.\n- **Dead ends are the most valuable content.** What was tried and did not work saves the resumer from spending the same hours. Record it under `Context`, with the reason it failed.\n- **Next steps are actionable and ordered** — the next thing to type, not a topic to consider.\n- **No secrets.** The file is committed. Never paste tokens, keys, credentials or `.env` values into it; reference where they live instead.\n- **No attribution.** No agent or model name, no session url, no \"handed off by\". The document says what the work is, never who typed it — see [Guardrails](#guardrails).\n\n### 5. Commit the handoff\n\nCommit the new file via `atomic-commit` and push. Until it is pushed, the handoff has not happened. Report the id, the path and the branch.\n\n**`atomic-commit` is optional** — at every commit this skill makes: the work at step 2, the handoff here, and the deletion at [step 8](#8-delete-the-handoff-when-the-work-is-done). Not installed, commit directly in the repo's own Conventional Commits conventions and push as usual. A handoff that never lands — or never clears — because a helper skill is absent is the one failure these steps cannot afford.\n\n### 6. Find the handoff to resume\n\n| Input                      | Action                                                                        |\n| :------------------------- | :---------------------------------------------------------------------------- |\n| An explicit id (`0003`)    | Open `.agents\u002Fhandoffs\u002F0003-*.md`. No match → say so, list what exists, stop. |\n| No id, exactly one handoff | That one. Unambiguous, so no question needed — name it in the report.         |\n| No id, several handoffs    | **List them and ask.** Never guess.                                           |\n| No id, none at all         | Say so and stop.                                                              |\n\n**Never resolve a handoff by recency.** \"The latest\" is a guess with a plausible-looking answer, and picking the wrong thread of work is the one failure a handoff cannot recover from. Two files sharing an id (a real race) → report the ambiguity, do not pick.\n\n### 7. Restore the ground, then work\n\nRead the document **fully** before touching anything. Then `git fetch`, check out the `branch` it names, and confirm the tree matches what `Progress` describes — if it does not, say so before working; the document is the writer's snapshot, and the branch may have moved since.\n\nThen do the work: `Goal` is the target, `Next steps` is the running order, `Context` is what you do not need to rediscover. `Open questions` are for the **human** — put them to the user rather than answering them by fiat.\n\n**Do not delete the handoff now.** Reading it is not consuming it.\n\n### 8. Delete the handoff when the work is done\n\nThe file is deleted **in the same commit that finishes the work** — not on read, not on resume, not on the first green test.\n\nA handoff deleted on read exists nowhere the moment the resuming session crashes — which is precisely the situation it was written to survive. Its lifetime is the work's lifetime: it lives while the work is unfinished and disappears with the last commit that finishes it. Nothing else in the repo records that it ever existed, and nothing needs to.\n\n### 9. Unfinished? Update in place\n\nResumed work that is being parked again **updates the existing document** — same id, same file. Refresh `Progress`, `Next steps`, `updated`, and `branch` if it changed; fold what you learned into `Context`. Never open a second handoff for the same thread of work: one thread, one id, until it is done.\n\n## Guardrails\n\n- **Reachable or honest.** Never write a handoff that implies work is available elsewhere when it was never pushed ([step 2](#2-make-the-work-reachable--before-writing-anything)).\n- **`atomic-commit` is optional, never a precondition.** It is the preferred way to commit the work, the handoff and the deletion that closes it, but if it is not installed, commit in the repo's own conventions and push anyway — a missing helper must not be why the handoff fails to land or fails to clear.\n- **Self-contained or worthless.** The reader has none of your session. Every reference is resolvable from the document alone.\n- **Delete on done, never on read** ([step 8](#8-delete-the-handoff-when-the-work-is-done)). The file survives every intermediate failure, by design.\n- **Never guess which handoff** ([step 6](#6-find-the-handoff-to-resume)). Explicit id, or the single unambiguous one, or ask.\n- **Ids are permanent and never reused** — even though the files are not ([REFERENCE.md](REFERENCE.md#id-allocation)).\n- **Attribution-free.** No agent or model name, no `Generated with`\u002F🤖 line, no session url, in the document, the commit, or anything this skill writes. The frontmatter has no author field on purpose — a handoff is a note to the next worker, not a signed artifact, and the next worker gains nothing from knowing which tool typed it.\n- **Secret-free.** The document is committed. Scan what you paste into `Context` — logs, env dumps and config excerpts are where secrets get in.\n- **A handoff is not an ADR.** Same `NNNN-title.md` shape, opposite lifecycle — an ADR (`write-docs`) is a permanent record that must never be deleted; a handoff is working state that is consumed and removed. Never file a decision as a handoff, and never let a handoff's deletion take a decision's record with it — if the work settled something architectural, that belongs in an ADR before the handoff goes ([REFERENCE.md](REFERENCE.md#relationship-to-the-siblings)).\n\n## Reference\n\n**Two branches, and step 1 settles which.** **Writing one** — open it at step 3 for the folder and file schema and how an id is allocated (and what happens when two runs allocate the same one), then at step 4 for the document contract and status vocabulary, since a document that quietly assumes the writer's session is the failure this skill exists to prevent. **Resuming one** — open it at step 6 for the resume rules and the lifecycle that puts the delete at the end of the work rather than at the read. And **when the note looks like it should be something else** — an ADR, a tracker issue, a commit — the same file draws that line: [REFERENCE.md](REFERENCE.md).\n","# handoff\n\nMove **in-progress work** from one agent to the next — a different session, a different machine, a different day — through a **structured document committed to the repo**. Two directions, one skill: **write** a handoff, or **resume** one.\n\nThe document is the whole product. It is written for a reader who has **none of the writer's session** — no scrollback, no open files, no memory of what was already tried. A handoff that only makes sense to the agent that wrote it has failed at the one job it has.\n\n## Workflow\n\n### 1. Route\n\n| Situation                                                             | Action                       |\n| :-------------------------------------------------------------------- | :--------------------------- |\n| Work is in flight and must be parked or passed on                     | **Write** — steps 2–5        |\n| A handoff exists and the user wants to continue it (`0003`, \"resume\") | **Resume** — steps 6–8       |\n| Work was resumed but is unfinished and is being parked again          | **Update in place** — step 9 |\n\n### 2. Make the work reachable — before writing anything\n\nA handoff is committed **so another machine can pick it up**. That promise is only kept if the _work_ travels too, not just the note about it.\n\n- **Uncommitted work is unreachable.** Commit it (via `atomic-commit`) and push the branch. Work in progress that is not yet coherent goes in one honest `wip` commit — the resumer needs the code more than it needs a tidy history.\n- **Nothing to commit** is a fine answer — say so in the document (`Progress` records \"nothing uncommitted\").\n- **Cannot push** (no remote, offline, work that must not leave the machine) → still write the handoff, but state it plainly under `Progress` and drop the `branch` field. A handoff describing work that exists nowhere but this disk is a **local note**, and must not claim otherwise.\n\n### 3. Allocate the id\n\nHandoffs live in **`.agents\u002Fhandoffs\u002F`**, named `NNNN-lower-kebab-case.md` (`0001-refactor-auth-layer.md`).\n\n**The next id is one above the highest ever used — read from git history, not the folder.** Handoffs are deleted when their work completes, so the folder is a poor witness; history is the only one that remembers `0003` existed. Ids are never reused, renumbered or gap-filled. Recipe: [REFERENCE.md](REFERENCE.md#id-allocation).\n\n### 4. Write the document\n\nCopy [`templates\u002Fhandoff.md`](templates\u002Fhandoff.md) and fill every section. Frontmatter carries the machine-readable fields (`title`, `status`, `created`, `updated`, `branch`, `issue`); the body carries the prose, in fixed H2s: `Goal`, `Context`, `Progress`, `Next steps`, `Open questions`. Full contract: [REFERENCE.md](REFERENCE.md#document-contract).\n\nWrite it **self-contained**:\n\n- **No deixis.** No \"the file we looked at\", \"as discussed\", \"the failing test from earlier\". Name paths, symbols, commands, issue numbers and commit shas outright.\n- **Dead ends are the most valuable content.** What was tried and did not work saves the resumer from spending the same hours. Record it under `Context`, with the reason it failed.\n- **Next steps are actionable and ordered** — the next thing to type, not a topic to consider.\n- **No secrets.** The file is committed. Never paste tokens, keys, credentials or `.env` values into it; reference where they live instead.\n- **No attribution.** No agent or model name, no session url, no \"handed off by\". The document says what the work is, never who typed it — see [Guardrails](#guardrails).\n\n### 5. Commit the handoff\n\nCommit the new file via `atomic-commit` and push. Until it is pushed, the handoff has not happened. Report the id, the path and the branch.\n\n**`atomic-commit` is optional** — at every commit this skill makes: the work at step 2, the handoff here, and the deletion at [step 8](#8-delete-the-handoff-when-the-work-is-done). Not installed, commit directly in the repo's own Conventional Commits conventions and push as usual. A handoff that never lands — or never clears — because a helper skill is absent is the one failure these steps cannot afford.\n\n### 6. Find the handoff to resume\n\n| Input                      | Action                                                                        |\n| :------------------------- | :---------------------------------------------------------------------------- |\n| An explicit id (`0003`)    | Open `.agents\u002Fhandoffs\u002F0003-*.md`. No match → say so, list what exists, stop. |\n| No id, exactly one handoff | That one. Unambiguous, so no question needed — name it in the report.         |\n| No id, several handoffs    | **List them and ask.** Never guess.                                           |\n| No id, none at all         | Say so and stop.                                                              |\n\n**Never resolve a handoff by recency.** \"The latest\" is a guess with a plausible-looking answer, and picking the wrong thread of work is the one failure a handoff cannot recover from. Two files sharing an id (a real race) → report the ambiguity, do not pick.\n\n### 7. Restore the ground, then work\n\nRead the document **fully** before touching anything. Then `git fetch`, check out the `branch` it names, and confirm the tree matches what `Progress` describes — if it does not, say so before working; the document is the writer's snapshot, and the branch may have moved since.\n\nThen do the work: `Goal` is the target, `Next steps` is the running order, `Context` is what you do not need to rediscover. `Open questions` are for the **human** — put them to the user rather than answering them by fiat.\n\n**Do not delete the handoff now.** Reading it is not consuming it.\n\n### 8. Delete the handoff when the work is done\n\nThe file is deleted **in the same commit that finishes the work** — not on read, not on resume, not on the first green test.\n\nA handoff deleted on read exists nowhere the moment the resuming session crashes — which is precisely the situation it was written to survive. Its lifetime is the work's lifetime: it lives while the work is unfinished and disappears with the last commit that finishes it. Nothing else in the repo records that it ever existed, and nothing needs to.\n\n### 9. Unfinished? Update in place\n\nResumed work that is being parked again **updates the existing document** — same id, same file. Refresh `Progress`, `Next steps`, `updated`, and `branch` if it changed; fold what you learned into `Context`. Never open a second handoff for the same thread of work: one thread, one id, until it is done.\n\n## Guardrails\n\n- **Reachable or honest.** Never write a handoff that implies work is available elsewhere when it was never pushed ([step 2](#2-make-the-work-reachable--before-writing-anything)).\n- **`atomic-commit` is optional, never a precondition.** It is the preferred way to commit the work, the handoff and the deletion that closes it, but if it is not installed, commit in the repo's own conventions and push anyway — a missing helper must not be why the handoff fails to land or fails to clear.\n- **Self-contained or worthless.** The reader has none of your session. Every reference is resolvable from the document alone.\n- **Delete on done, never on read** ([step 8](#8-delete-the-handoff-when-the-work-is-done)). The file survives every intermediate failure, by design.\n- **Never guess which handoff** ([step 6](#6-find-the-handoff-to-resume)). Explicit id, or the single unambiguous one, or ask.\n- **Ids are permanent and never reused** — even though the files are not ([REFERENCE.md](REFERENCE.md#id-allocation)).\n- **Attribution-free.** No agent or model name, no `Generated with`\u002F🤖 line, no session url, in the document, the commit, or anything this skill writes. The frontmatter has no author field on purpose — a handoff is a note to the next worker, not a signed artifact, and the next worker gains nothing from knowing which tool typed it.\n- **Secret-free.** The document is committed. Scan what you paste into `Context` — logs, env dumps and config excerpts are where secrets get in.\n- **A handoff is not an ADR.** Same `NNNN-title.md` shape, opposite lifecycle — an ADR (`write-docs`) is a permanent record that must never be deleted; a handoff is working state that is consumed and removed. Never file a decision as a handoff, and never let a handoff's deletion take a decision's record with it — if the work settled something architectural, that belongs in an ADR before the handoff goes ([REFERENCE.md](REFERENCE.md#relationship-to-the-siblings)).\n\n## Reference\n\n**Two branches, and step 1 settles which.** **Writing one** — open it at step 3 for the folder and file schema and how an id is allocated (and what happens when two runs allocate the same one), then at step 4 for the document contract and status vocabulary, since a document that quietly assumes the writer's session is the failure this skill exists to prevent. **Resuming one** — open it at step 6 for the resume rules and the lifecycle that puts the delete at the end of the work rather than at the read. And **when the note looks like it should be something else** — an ADR, a tracker issue, a commit — the same file draws that line: [REFERENCE.md](REFERENCE.md).\n",{"name":5,"metadata":24,"description":7,"allowed-tools":26},{"summary":25},"Hands off in-progress work to another agent or session via a committed handoff document.",[27,28,29,30,31,32,33,34,35,36,37],"Read","Write","Edit","Grep","Glob","Bash(jq:*)","Bash(printf:*)","Bash(git log:*)","Bash(git rev-parse:*)","Bash(ls:*)","Bash(tail:*)","main","0343b3844cdfe00ebabe556f2a63c0515208e92c","skills\u002Fwork\u002Fhandoff",0,"https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills","https:\u002F\u002Fgithub.com\u002FTitusKirch","https:\u002F\u002Fkirch.dev\u002F","💼 Self-employed Web Developer\r\n🎓 B. Sc. Computer Science","2026-11-02 00:46:17",{"source":8,"skillName":5,"commands":48,"prompt":58,"zipAvailable":59},[49,52,55],{"tool":50,"command":51},"npx","npx skills add TitusKirch\u002Fskills --skill handoff",{"tool":53,"command":54},"bunx","bunx skills add TitusKirch\u002Fskills --skill handoff",{"tool":56,"command":57},"pnpm","pnpm dlx skills add TitusKirch\u002Fskills --skill handoff","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fhandoff\u002FSKILL.md",true,[61,67,71],{"path":62,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":65,"sourceHtmlUrl":66},"REFERENCE.md","REFERENCE","参考文档",21365,"https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fhandoff\u002FREFERENCE.md",{"path":68,"resourceType":63,"resourceTypeDesc":64,"sizeBytes":69,"sourceHtmlUrl":70},"templates\u002Fhandoff.md",1867,"https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fhandoff\u002Ftemplates\u002Fhandoff.md",{"path":72,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":75,"sourceHtmlUrl":76},"templates\u002Fresolve-config.sh","SCRIPT","脚本",2402,"https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fhandoff\u002Ftemplates\u002Fresolve-config.sh",[78,85,92,99,106,113],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":82,"categories":83,"keywords":84,"publishedAt":19},"01M74FV6SYWA6MEBCX3EY64484","gamma","The third fixture skill — carries the worklock, plan and tldr blocks, and names the resolver without hosting the config block.","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Fregistry\u002Fskills\u002Fwork\u002Fgamma\u002FSKILL.md",[],[],{"skillNo":86,"name":87,"nameZh":6,"description":88,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":19},"01M74FV4K7PNY8GTJH2M8ZTJGT","beta","The second fixture skill — carries the tagged blocks but no config block.","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Fregistry\u002Fskills\u002Frepo\u002Fbeta\u002FSKILL.md",[],[],{"skillNo":93,"name":94,"nameZh":6,"description":95,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":96,"categories":97,"keywords":98,"publishedAt":19},"01M74FV2JYBQ3VRP180AZEVQFC","alpha","The first fixture skill — carries the config block and the resolver.","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Fregistry\u002Fskills\u002Frepo\u002Falpha\u002FSKILL.md",[],[],{"skillNo":100,"name":101,"nameZh":6,"description":102,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":19},"01M74FTVVHVDTMTP0EX4W924NK","work-review","Reviews a single tracked issue's pushed implementation across GitHub (gh), GitLab (glab), Linear (MCP) or local issue files as a fresh, independent agent — reads the issue's requirements and the pushed diff, adversarially checks whether the work is correct and complete, and writes a verdict that routes the issue to done (accepted), changes-requested (feedback, back to the implement loop), needs-human (escalation), or blocked. It never implements or fixes anything — review only. It also applies a human's verdict on a needs-human issue. Tracker, labels and the round cap come from the committed config (.tituskirch-skills.json). Use when the user wants to review, check, or sign off one specific issue's AI work, mentions an issue awaiting review, or says things like \"review issue 42\", \"check the work on issue 42\", \"reviewe Issue 42\", or gives a verdict on an escalated issue.","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fwork-review\u002FSKILL.md",[],[],{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":19},"01M74FTNQ1RCJTZ929GNJ2FBX5","work-review-queue","Drains a repo's queue of issues awaiting AI review across GitHub (gh), GitLab (glab), Linear (MCP) or local issue files — selects every issue in review, then reviews each with a fresh, independent agent by delegating to work-review, routing each to done, changes-requested (back to the implement loop), needs-human, or blocked. Starts by reconciling issues whose PR a human merged or closed out-of-band. Honours a per-run cap, single-flight-locked with a lock separate from the implement loop's, so review and implement drains run concurrently. Use when the user wants to review, drain, or auto-review the pushed AI work, run the review loop, says things like \"review the queue\", \"reviewe die Issues\", \"drain the review queue\", or starts a repeating driver.","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fwork-review-queue\u002FSKILL.md",[],[],{"skillNo":114,"name":115,"nameZh":6,"description":116,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":19},"01M74FTFKF2GBCKMKKHN5QJW0G","work-implement","Implements a single tracked issue across GitHub (gh), GitLab (glab), Linear (MCP) or local issue files — claims it via the lifecycle label, implements on a branch (fresh from ready, or re-work from changes-requested after review feedback), runs the repo's checks, commits and PUSHES, and hands the issue to the review loop by advancing the label to review. It never reviews its own work — the separate work-review skill does that. Tracker, label lifecycle and branch strategy come from the committed config (.tituskirch-skills.json). Use when the user wants to work, implement, action or pick up one specific issue or ticket, mentions an ai-ready issue, or says things like \"work issue 42\", \"arbeite Issue 42 ab\", \"implementiere Ticket X\", or \"address the review feedback on issue Y\".","https:\u002F\u002Fgithub.com\u002FTitusKirch\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fwork\u002Fwork-implement\u002FSKILL.md",[],[]]