[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79234W72RKWYV2S38G2B0SY":3,"skill-same:01M79234W72RKWYV2S38G2B0SY":48},{"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":29,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M79234W72RKWYV2S38G2B0SY","plan-with-me",null,"Collaborative, human-in-the-loop planning for something new that spans several parts and needs human judgement at each fork. A brainstorming partner that funnels — each fork opens up the real options, then narrows to a decision, surfacing caveats and breaking points, and co-authoring the plan instead of guessing alone. Fits any domain where you plan something new against existing context. Resolves one decision at a time, each with a recommendation, then emits a lightweight plan to hand off for implementation. Use when planning alone would make wrong assumptions, or the user wants to shape something new together.","dotlas\u002Fskills","dotlas","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F88832003?v=4","01M791ZCS9BAQ40PZ8ADG8TCR1","Dotlas","01M791ZCT18PGXY969S8EZ1MVV",2,"MIT","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fplan-with-me\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: plan-with-me\ndescription: \"Collaborative, human-in-the-loop planning for something new that spans several parts and needs human judgement at each fork. A brainstorming partner that funnels — each fork opens up the real options, then narrows to a decision, surfacing caveats and breaking points, and co-authoring the plan instead of guessing alone. Fits any domain where you plan something new against existing context. Resolves one decision at a time, each with a recommendation, then emits a lightweight plan to hand off for implementation. Use when planning alone would make wrong assumptions, or the user wants to shape something new together.\"\n---\n# Plan With Me — Collaborative Planning\n\nA brainstorming partner for building something new.\nIt **funnels**: each fork opens up to weigh the real options, then narrows to a decision\n— surfacing the caveats and breaking points before they cost anything.\nYou stay in the decision seat the whole way — the plan is co-authored, not handed down.\n\nThe reason to keep a human in the loop: an agent executes a plan well, but it can’t know\nwhich tradeoff you prefer when several options are valid, can’t see the context that\nisn’t written down (“our clients hate X”, “this platform can’t do Y”), can’t judge scope\n(v1 requirement, or defer?), and won’t notice when a design choice quietly contradicts a\nconstraint you never stated.\nSo every real fork passes through you, and the result reflects your actual intent rather\nthan the agent’s best guess at it.\n\nThe output is deliberately **lightweight** — a vision, a decision log, and the open\nquestions.\nIt’s raw material for implementation or a hardening pass (e.g. `closed-plan`),\nnot the final build spec.\nScope scales to the work: a ten-stage plan for a new app, or a single page for a tweak\nto something that already exists.\n\n* * *\n\n## Posture\n\n- **Collaborative, not adversarial.** Challenge is woven in gently — options offered,\n  not objections thrown.\n  This isn’t interrogating a finished plan; it’s building one together.\n- **Plans, not code.** This produces a plan.\n  Running a command to check feasibility is fine; writing the implementation is not.\n- **Match the ambition to the work.** Don’t force a one-line tweak through ten stages,\n  or compress a new app into a paragraph.\n- **A conversation, not a questionnaire.** Forks are talked through in text as much as\n  picked in a prompt. `AskUserQuestion` locks a decision — it doesn’t replace the\n  discussion that earns it.\n\n* * *\n\n## The process\n\n### Phase 0 — Establish context of what already exists\n\nBefore engaging with the idea, inspect the current context so you know what is genuinely\n*new* versus what to reuse or extend.\nDon’t assume conventions — find them.\n\n*In a codebase:* read the structure and the rules it already follows — root files and\nmonorepo indicators, instruction and architecture docs (`.cursor\u002F`, `.agents\u002F`,\n`CONTRIBUTING.md`, `ARCHITECTURE.md`, root `*.md`), package and build layout,\nlint\u002Ftest\u002FCI config, and any existing feature similar to what’s being planned (staging\ntables, proposal flows, queue mechanisms) that the new work should follow or extend.\n\n*In other work:* the equivalent existing material — prior documents and house style, an\nexisting design system with its components and assets, an established process or\nprecedent.\n\nThen tell the user briefly what you found (“the repo uses X, has conventions for Y, and\nthere’s an existing Z we could build on”). That grounds everything after it.\nIf nothing is documented, say so and proceed on general best practice — don’t invent\nconventions that aren’t there.\n\nAlso note which plan directory the project uses (`.plan\u002F`, `.plans\u002F`, `plans\u002F`) — fall\nback to `.plan\u002F` if none exists.\nPlans will be saved there.\n\n### Phase 1 — Understand the vision\n\nListen to the idea. Ask clarifying questions only if the core intent is genuinely\nambiguous. Then summarise it back in 2-3 sentences, decompose it into its separate\nconcerns, and pick the one to resolve first — usually the foundation everything else\ndepends on (in code, often the shape of the data).\n\n### Phase 2 — Funnel, one decision at a time\n\nThis is the core. Work through the concerns in dependency order — foundational decisions\nbefore the ones that build on them — resolving each through an adaptive\nopen→discuss→lock conversation.\nThe through-line is convergence: the plan gets narrower and firmer with each fork; the\ndivergence is there to make each convergence well-founded, not to keep options open for\ntheir own sake.\n\nFor each fork, follow the pattern:\n\n**1. Size the fork.** Judge how far to open it up.\nGo deep when the decision is high stakes or hard to reverse, when several genuinely\nviable or non-obvious options exist, or when the user is exploring or uncertain.\nKeep it quick when the decision is local, cheaply changed, has an obvious default, or\nthe user is decisive.\n\n**2. Open it up (when it warrants).** Lay the option space out in plain text — candidate\napproaches including the non-obvious and hybrid ones, what each buys and costs, relevant\ninformation, prior art, a concrete example or two to draw inspiration from.\nLook under every rock so neither party discovers a better path *after* the decision is\nlocked.\n\n**3. Talk it through.** Let it be a back-and-forth — the user reacts, adds context you\ncouldn’t have known, rules things out, points somewhere new; you refine the framing in\nresponse. Free text, not a form.\nMost of the real thinking happens here.\n\n**4. Lock it.** Converge when the space feels mapped — use your judgement on timing;\nthere is no fixed number of exchanges.\nIf the discussion already produced a clear answer, state it back and confirm in text.\nIf a crisp pick between comparable options remains, lock it with `AskUserQuestion`.\n\n**When you lock with `AskUserQuestion`** — one call, one question (never batch multiple\ndecisions into a single call, as each answer shapes the next question):\n- `question`: the specific decision to make, stated plainly.\n- `header`: a short tag ≤ 12 chars for the decision topic (e.g. `”Data shape”`,\n  `”Auth”`).\n- `options`: 2-4 concrete choices — not an open-ended prompt.\n  Place the recommended option first and append `(Recommended)` to its label; put the\n  reasoning in that option’s `description`. Put counter-arguments or trade-offs in the\n  other options’ descriptions.\n- `multiSelect: false` (decisions are single-choice).\n- An “Other” option is provided automatically — the user is never boxed in.\n\nAs decisions accumulate, keep a **running decision log** — numbered, each capturing what\nwas decided, the options considered, why this one won, and any constraint it imposes\ndownstream. Cite earlier decisions by number (“since #4 locked us into X, we’re limited\nto Y or Z here”).\n\nSharpen the funnel as you go:\n- **Test assumptions against reality before baking them in.** When a choice depends on\n  whether something is actually possible, check rather than assume — *in code,* run a\n  command (does the database support this extension?\n  does the endpoint accept this format?); *elsewhere,* verify against the real\n  constraint (a statute, a spec, whether the design system has that component).\n  Surface blockers early, not after the plan is written.\n- **Use real examples to stress-test.** When the user offers concrete input — sample\n  data, a screenshot, an actual document — run the emerging design against it: “given\n  this real case, what would we produce?\n  does the model handle it?”\n  Concrete examples expose gaps that abstract discussion glides over.\n- **Push back when it helps** — a simpler alternative exists, two requirements are in\n  tension, an existing pattern already solves part of it, or scope is outgrowing what’s\n  buildable in reasonable stages.\n  Frame it as an option: “that works, but X would simplify Y at the cost of Z.”\n\nAdapt to how the user responds:\n- Gives a real example → pause and validate against it.\n- Says “you decide” → make the call, state your reasoning, move on.\n- Wants to defer → note it as an open question and continue with what’s unblocked.\n- Goes on a tangent → capture anything useful as a note, then redirect.\n- Asks something the context already answers → answer it from Phase 0 or a quick check;\n  don’t hand it back to them.\n\n### Phase 3 — Emit the plan\n\nWhen the critical decisions are settled (minor ones always remain — that’s fine), write\na lightweight plan and save it to the plan directory detected in Phase 0. Filename:\n`YYYY-MM-DD-{topic-slug}.md` (kebab-cased from the Phase 1 vision).\nTell the user the path.\nIt contains:\n- **Vision** — the feature from first principles, the why and the what, so anyone\n  picking up a piece understands the whole.\n- **Decision log** — the numbered decisions from the session.\n- **Terminology** — any new domain terms coined along the way.\n- **Key constraints** — the rules the build must honour.\n- **Open questions** — deferred decisions to resolve during implementation.\n\nKeep it to the material and the intent.\nDeterministic sequencing, step-by-step staging, and convention-compliance belong to the\nimplementation or hardening pass this feeds into — don’t duplicate that work here.\nFor a large effort you can split the plan into dependency-ordered stages; for a small\none, a single page is the honest size.\n\n* * *\n\n## If the session pauses\n\nOn “save progress” or “let’s pause”, write the current state — decisions, open\nquestions, notes — to the plan file (same path as Phase 3, overwriting any prior\ncheckpoint). Mark it `status: in-progress` at the top; flip to `status: complete` when\nPhase 3 finalises it.\nTell the user the path.\nResuming is then just reading it back.\n","# Plan With Me — Collaborative Planning\n\nA brainstorming partner for building something new.\nIt **funnels**: each fork opens up to weigh the real options, then narrows to a decision\n— surfacing the caveats and breaking points before they cost anything.\nYou stay in the decision seat the whole way — the plan is co-authored, not handed down.\n\nThe reason to keep a human in the loop: an agent executes a plan well, but it can’t know\nwhich tradeoff you prefer when several options are valid, can’t see the context that\nisn’t written down (“our clients hate X”, “this platform can’t do Y”), can’t judge scope\n(v1 requirement, or defer?), and won’t notice when a design choice quietly contradicts a\nconstraint you never stated.\nSo every real fork passes through you, and the result reflects your actual intent rather\nthan the agent’s best guess at it.\n\nThe output is deliberately **lightweight** — a vision, a decision log, and the open\nquestions.\nIt’s raw material for implementation or a hardening pass (e.g. `closed-plan`),\nnot the final build spec.\nScope scales to the work: a ten-stage plan for a new app, or a single page for a tweak\nto something that already exists.\n\n* * *\n\n## Posture\n\n- **Collaborative, not adversarial.** Challenge is woven in gently — options offered,\n  not objections thrown.\n  This isn’t interrogating a finished plan; it’s building one together.\n- **Plans, not code.** This produces a plan.\n  Running a command to check feasibility is fine; writing the implementation is not.\n- **Match the ambition to the work.** Don’t force a one-line tweak through ten stages,\n  or compress a new app into a paragraph.\n- **A conversation, not a questionnaire.** Forks are talked through in text as much as\n  picked in a prompt. `AskUserQuestion` locks a decision — it doesn’t replace the\n  discussion that earns it.\n\n* * *\n\n## The process\n\n### Phase 0 — Establish context of what already exists\n\nBefore engaging with the idea, inspect the current context so you know what is genuinely\n*new* versus what to reuse or extend.\nDon’t assume conventions — find them.\n\n*In a codebase:* read the structure and the rules it already follows — root files and\nmonorepo indicators, instruction and architecture docs (`.cursor\u002F`, `.agents\u002F`,\n`CONTRIBUTING.md`, `ARCHITECTURE.md`, root `*.md`), package and build layout,\nlint\u002Ftest\u002FCI config, and any existing feature similar to what’s being planned (staging\ntables, proposal flows, queue mechanisms) that the new work should follow or extend.\n\n*In other work:* the equivalent existing material — prior documents and house style, an\nexisting design system with its components and assets, an established process or\nprecedent.\n\nThen tell the user briefly what you found (“the repo uses X, has conventions for Y, and\nthere’s an existing Z we could build on”). That grounds everything after it.\nIf nothing is documented, say so and proceed on general best practice — don’t invent\nconventions that aren’t there.\n\nAlso note which plan directory the project uses (`.plan\u002F`, `.plans\u002F`, `plans\u002F`) — fall\nback to `.plan\u002F` if none exists.\nPlans will be saved there.\n\n### Phase 1 — Understand the vision\n\nListen to the idea. Ask clarifying questions only if the core intent is genuinely\nambiguous. Then summarise it back in 2-3 sentences, decompose it into its separate\nconcerns, and pick the one to resolve first — usually the foundation everything else\ndepends on (in code, often the shape of the data).\n\n### Phase 2 — Funnel, one decision at a time\n\nThis is the core. Work through the concerns in dependency order — foundational decisions\nbefore the ones that build on them — resolving each through an adaptive\nopen→discuss→lock conversation.\nThe through-line is convergence: the plan gets narrower and firmer with each fork; the\ndivergence is there to make each convergence well-founded, not to keep options open for\ntheir own sake.\n\nFor each fork, follow the pattern:\n\n**1. Size the fork.** Judge how far to open it up.\nGo deep when the decision is high stakes or hard to reverse, when several genuinely\nviable or non-obvious options exist, or when the user is exploring or uncertain.\nKeep it quick when the decision is local, cheaply changed, has an obvious default, or\nthe user is decisive.\n\n**2. Open it up (when it warrants).** Lay the option space out in plain text — candidate\napproaches including the non-obvious and hybrid ones, what each buys and costs, relevant\ninformation, prior art, a concrete example or two to draw inspiration from.\nLook under every rock so neither party discovers a better path *after* the decision is\nlocked.\n\n**3. Talk it through.** Let it be a back-and-forth — the user reacts, adds context you\ncouldn’t have known, rules things out, points somewhere new; you refine the framing in\nresponse. Free text, not a form.\nMost of the real thinking happens here.\n\n**4. Lock it.** Converge when the space feels mapped — use your judgement on timing;\nthere is no fixed number of exchanges.\nIf the discussion already produced a clear answer, state it back and confirm in text.\nIf a crisp pick between comparable options remains, lock it with `AskUserQuestion`.\n\n**When you lock with `AskUserQuestion`** — one call, one question (never batch multiple\ndecisions into a single call, as each answer shapes the next question):\n- `question`: the specific decision to make, stated plainly.\n- `header`: a short tag ≤ 12 chars for the decision topic (e.g. `”Data shape”`,\n  `”Auth”`).\n- `options`: 2-4 concrete choices — not an open-ended prompt.\n  Place the recommended option first and append `(Recommended)` to its label; put the\n  reasoning in that option’s `description`. Put counter-arguments or trade-offs in the\n  other options’ descriptions.\n- `multiSelect: false` (decisions are single-choice).\n- An “Other” option is provided automatically — the user is never boxed in.\n\nAs decisions accumulate, keep a **running decision log** — numbered, each capturing what\nwas decided, the options considered, why this one won, and any constraint it imposes\ndownstream. Cite earlier decisions by number (“since #4 locked us into X, we’re limited\nto Y or Z here”).\n\nSharpen the funnel as you go:\n- **Test assumptions against reality before baking them in.** When a choice depends on\n  whether something is actually possible, check rather than assume — *in code,* run a\n  command (does the database support this extension?\n  does the endpoint accept this format?); *elsewhere,* verify against the real\n  constraint (a statute, a spec, whether the design system has that component).\n  Surface blockers early, not after the plan is written.\n- **Use real examples to stress-test.** When the user offers concrete input — sample\n  data, a screenshot, an actual document — run the emerging design against it: “given\n  this real case, what would we produce?\n  does the model handle it?”\n  Concrete examples expose gaps that abstract discussion glides over.\n- **Push back when it helps** — a simpler alternative exists, two requirements are in\n  tension, an existing pattern already solves part of it, or scope is outgrowing what’s\n  buildable in reasonable stages.\n  Frame it as an option: “that works, but X would simplify Y at the cost of Z.”\n\nAdapt to how the user responds:\n- Gives a real example → pause and validate against it.\n- Says “you decide” → make the call, state your reasoning, move on.\n- Wants to defer → note it as an open question and continue with what’s unblocked.\n- Goes on a tangent → capture anything useful as a note, then redirect.\n- Asks something the context already answers → answer it from Phase 0 or a quick check;\n  don’t hand it back to them.\n\n### Phase 3 — Emit the plan\n\nWhen the critical decisions are settled (minor ones always remain — that’s fine), write\na lightweight plan and save it to the plan directory detected in Phase 0. Filename:\n`YYYY-MM-DD-{topic-slug}.md` (kebab-cased from the Phase 1 vision).\nTell the user the path.\nIt contains:\n- **Vision** — the feature from first principles, the why and the what, so anyone\n  picking up a piece understands the whole.\n- **Decision log** — the numbered decisions from the session.\n- **Terminology** — any new domain terms coined along the way.\n- **Key constraints** — the rules the build must honour.\n- **Open questions** — deferred decisions to resolve during implementation.\n\nKeep it to the material and the intent.\nDeterministic sequencing, step-by-step staging, and convention-compliance belong to the\nimplementation or hardening pass this feeds into — don’t duplicate that work here.\nFor a large effort you can split the plan into dependency-ordered stages; for a small\none, a single page is the honest size.\n\n* * *\n\n## If the session pauses\n\nOn “save progress” or “let’s pause”, write the current state — decisions, open\nquestions, notes — to the plan file (same path as Phase 3, overwriting any prior\ncheckpoint). Mark it `status: in-progress` at the top; flip to `status: complete` when\nPhase 3 finalises it.\nTell the user the path.\nResuming is then just reading it back.\n",{"name":5,"description":7},"main","44a7c5d185b1020f689d2129547d7470c3715e76","skills\u002Fcustom\u002Fcommon\u002Fplan-with-me",0,"https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills","https:\u002F\u002Fwww.skills.sh\u002Fdotlas\u002Fskills","https:\u002F\u002Fgithub.com\u002Fdotlas","www.dotlas.com","A platform to help companies make better city-level decisions and improve the bottom line of their retail footprint.","2026-10-27 22:40:58",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add dotlas\u002Fskills --skill plan-with-me",{"tool":40,"command":41},"bunx","bunx skills add dotlas\u002Fskills --skill plan-with-me",{"tool":43,"command":44},"pnpm","pnpm dlx skills add dotlas\u002Fskills --skill plan-with-me","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fplan-with-me\u002FSKILL.md",true,[],[49,56,63,70,77,84],{"skillNo":50,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":19},"01M792400972AXPY3XESVHRFRA","where-were-we","Provide the user with a refresher of existing chat turns. Useful when the user wants to understand where we stand in the conversation.","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fwhere-were-we\u002FSKILL.md",[],[],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7923T58GM9G9TVAZB9TGTWE","update-global-dotlas-skills","Sync the globally-installed dotlas\u002Fskills catalog and prune stale remnants. Runs the documented global install command, then detects skills that were installed from dotlas\u002Fskills but no longer exist in the catalog and removes them (with confirmation). Use when the user wants to update, refresh, or clean up their global Dotlas skills, or invokes \u002Fupdate-global-dotlas-skills.","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fupdate-global-dotlas-skills\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7923KGCMXT25Y7XVC6G1ERX","unpack","Explain an unfamiliar thing plainly and from first principles, structured so the reader sees why it is true. Use when someone is trying to understand something — a dense or jargon-heavy passage, a baffling error, an unfamiliar mechanism, a codebase, or a document — asks to break something down (\"walk me through this\", \"what does this actually mean\"), or wants their intent turned into a precise instruction for another agent. Trigger on confusion or a request to understand, not on difficulty alone; technical or long content is not itself a signal.","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Funpack\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},"01M7923HNVVHRKRVQ5M0P7TXZA","sum","Use when the previous response is too technical, complex, or jargon-heavy and needs a clear, plain-language explanation","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fsum\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},"01M7923CY3FR71VPKBYRHZNNVH","slack","Compose a Slack message, or digest a Slack thread into compact context. Use when the user asks to reply to a person or channel, write a Slack post, or catch up on a thread without reading the whole transcript. This skill covers how to write well for Slack — both jobs share the same standard.","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fslack\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},"01M7922ZHHS3XB2YBBD92BYPWB","multi-agent-chat","Open a chatroom — a plain .txt log on disk — so agents in separate sessions (Claude, Codex, others) whose work has collided or become correlated can talk to each other: resolve the conflict, trade the context each is missing, and settle who owns what. The agent that opens the room moderates it; the operator relays the room path to the other agents. Use when two sessions are stepping on each other or one needs what the other knows, or when handed a room path and asked to join.","https:\u002F\u002Fgithub.com\u002Fdotlas\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcustom\u002Fcommon\u002Fmulti-agent-chat\u002FSKILL.md",[],[]]