[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75C9F8PK76Z3FN2NZWD0S0T":3,"skill-same:01M75C9F8PK76Z3FN2NZWD0S0T":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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":31,"ownerHtmlUrl":32,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":33,"install":34,"resources":46},"01M75C9F8PK76Z3FN2NZWD0S0T","plandesk-foreman",null,"Runs the Plan Desk board floor — takes one task or a whole frontier of todos, grooms each into a build contract, cuts slices, dispatches implementation to worker CLIs, verifies their claims, gates each slice by its risk lane, and commits what clears. Use whenever asked to work a task, ship a ticket, take the next task, clear the todos, run the board, or hand implementation to a worker — even when the factory is not named.","asyncdotengineering\u002Fplandesk","asyncdotengineering","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F130723569?v=4","01M75C95Y312DEGHZMVT0W3FPB","Asyncdot Engineering","01M75C95Z7AWGPJCAY38XYPX0W",16,"MIT","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fplandesk-foreman\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: plandesk-foreman\ndescription: \"Runs the Plan Desk board floor — takes one task or a whole frontier of todos, grooms each into a build contract, cuts slices, dispatches implementation to worker CLIs, verifies their claims, gates each slice by its risk lane, and commits what clears. Use whenever asked to work a task, ship a ticket, take the next task, clear the todos, run the board, or hand implementation to a worker — even when the factory is not named.\"\nuser-invocable: true\nargument-hint: \"\u003Ctask id | 'next' | 'all todo' | goal name> [--to \u003Cworker>]\"\n---\n\n# Foreman\n\nTakes board work to committed. Planning fills the board, the foreman runs the floor,\nworkers build, a human decides what merges.\n\nThe policy this runs on lives beside it and is not repeated here — the cycle\ncontract in [factory.md](..\u002F..\u002Ffactory\u002Ffactory.md), dispatch and verification in\n[protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md), worker selection in\n[routing.md](..\u002F..\u002Ffactory\u002Frouting.md), gates in\n[lanes.md](..\u002F..\u002Ffactory\u002Flanes.md), slice shapes in\n[slicing.md](..\u002F..\u002Ffactory\u002Fslicing.md), the brief's additions in\n[brief.md](..\u002F..\u002Ffactory\u002Fbrief.md), the worker's standard in\n[workmanship.md](..\u002F..\u002Ffactory\u002Fworkmanship.md). Read the ones a run touches.\n\n## Workflow\n\n1. **Preflight.** Confirm the tree is clean, no dispatch is already running\n   (`pgrep -f`), the board answers, and at least one worker in\n   [workers\u002F](..\u002F..\u002Ffactory\u002Fworkers\u002F) passes its probe. Each of these turns a\n   long run into wasted tokens if it fails halfway, so check before starting,\n   not on the way past. Open the run with `start_agent_run`.\n\n2. **Resolve the scope.** A named task label (looked up with `list_tasks` \u002F\n   `get_task`) works one item. `next` calls\n   `get_next_task`. `all todo` or a goal name takes the whole frontier — only\n   `todo` tasks whose prerequisites are `done`; `scope` and `backlog` are\n   waiting on a human and are not yours to release. Read each task's linked\n   spec before judging it.\n\n3. **Groom inline — do not dispatch this.** Judge each task against the\n   Definition of Ready in [groom-task](..\u002Fplandesk-groom-task\u002FSKILL.md) and rewrite\n   anything short of it with `update_task` yourself, following that skill's\n   procedure — it is the one readiness bar this project keeps, so do not invent\n   a second one here. Grooming is judgment about intent, and shipping that\n   judgment to a worker is how a plan drifts from what was actually wanted. If a\n   task cannot be groomed without a decision you do not own, return it to\n   `scope` with a comment naming the decision, and carry on with the rest.\n\n   Grooming as a prelude to dispatch does **not** wait on groom-task's own\n   `approve` gate: post the groom comment (what changed, what was inferred) for\n   traceability, then proceed — the rewrite is reviewed together with the work\n   it produced, at the item's own lane gate, where a human can still reject\n   both. A groom left uncommented is still the violation.\n\n4. **Cut slices when the frontier is wider than one item.** One task needs no\n   slicing. Several become deliverable units per\n   [slicing.md](..\u002F..\u002Ffactory\u002Fslicing.md) — complete paths through the layers\n   they touch, sized to one worker's context, grouped so parallel workers do not\n   edit the same lines. The grouping goes in each brief's WBS snapshot, not in a\n   file of its own — the task descriptions already hold the build contract, so\n   do not restate the work anywhere.\n\n5. **Dispatch implementation — default async.** One dispatch per tree, per\n   [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md). Pick the worker from\n   [routing.md](..\u002F..\u002Ffactory\u002Frouting.md) unless one was named; a named worker\n   wins, and several named workers split the slices across worktrees. Build the\n   brief from [brief-template.md](..\u002F..\u002Ffactory\u002Fbrief-template.md) — copy,\n   substitute, paste workmanship in full — which carries protocol.md's\n   five-section contract: the bar, the result contract, the ground, the WBS\n   snapshot, and a live `Context:` link.\n\n   **Fire async unless the slice is trivial.** Background is the default for\n   every real dispatch. Run foreground only when the slice is genuinely\n   small (one obvious edit, expected under a few minutes) and the user asked\n   for inline work. The recipe:\n\n   1. Reset the task's `.agents\u002Ffactory\u002Fruns\u002F` state and capture the baseline per\n      [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md) dispatch step 2 — a stale\n      result file fires the monitor instantly, and verification needs the\n      pre-dispatch test counts to diff against.\n   2. Run the worker's `probe`; substitute `{prompt_file}` and `{repo_path}` in\n      its `command` template — flags come from [workers\u002F](..\u002F..\u002Ffactory\u002Fworkers\u002F),\n      never from memory.\n   3. Append the log redirect the engine owns:\n      `> .agents\u002Ffactory\u002Fruns\u002Fworker-\u003Ctask>.log 2>&1` (never put redirects in worker files).\n   4. **Background through the harness** — `run_in_background: true` on the\n      Shell\u002FBash tool. Never append `&` or wrap in `nohup … &`; that\n      orphan-detaches, the harness fires a false \"completed\", and the real leaf\n      keeps writing to a tree nobody is watching.\n   5. **Arm the monitor in the same step**, also backgrounded — the watch loop\n      in protocol.md's *Watching a live dispatch*. Do not dispatch and \"remember\n      to watch later.\"\n   6. **Do not poll the harness completion notification.** Wrapper exit is\n      unreliable (orphan shell, transient API blip). The monitor watches\n      `.agents\u002Ffactory\u002Fruns\u002Fresult-\u003Ctask>.json`; use `Await` on the monitor shell when you are\n      ready to verify, or continue other foreman work while it runs.\n\n   **Stdin is per worker** — getting this wrong backgrounds a hang:\n\n   | delivery | workers | rule |\n   | --- | --- | --- |\n   | prompt via stdin | claude, cursor, opencode | `\u003C {prompt_file}` — **no** `\u003C \u002Fdev\u002Fnull` |\n   | prompt via `@file` or arg | pi, codex, grok | pi: `@{prompt_file}`; codex: `\"$(cat …)\"`; grok: `--prompt-file` — add `\u003C \u002Fdev\u002Fnull` |\n\n   Mint the live link explicitly: `create_share_link` on the task (or the goal,\n   for a multi-slice run), `expires: 7d`, and put the returned `markdown_url`\n   in the brief as `Context:`. The worker has no MCP access, so this is the\n   only way it reads live board state instead of a copy that goes stale the\n   moment someone edits the task. Derive the WBS snapshot from `list_tasks` +\n   `list_edges` so the worker sees the agreed order and the paths a later item\n   owns; a worker given one node and no map finishes the next one too.\n\n   **Never dispatch a `kind: 'decision'` task.** Its resolution is a\n   conversation with whoever owns the outcome; a worker given one will answer\n   its own question and record a fabrication as a decision. Report it by task\n   **label** as awaiting a human and take the next frontier item — a decision\n   task blocks only itself; unlike a lane gate it does not stop the run.\n\n6. **Stage the moment a worker returns, before reading anything.** Review takes\n   minutes and unstaged work is defenceless for all of them; staged work\n   survives a later `git checkout` because git restores it from the index. This\n   ordering is the cheapest real protection in the whole run.\n\n   **Unless the dispatch was killed.** No result file means it ran no gates and\n   wrote no result, so the tree holds unverified partial output rather than work\n   product. Discard it and re-dispatch clean per\n   [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md) — staging it makes salvage look like\n   a deliverable, and the next reader cannot tell the difference.\n\n7. **Verify the claims.** Follow the verification sequence in\n   [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md) — gate integrity before re-running\n   anything, then suppressions, then per-package test counts against the\n   pre-dispatch baseline, then debris. Exit codes decide, not the worker's\n   summary. A dispatch that fails verification is recorded and re-scoped, not\n   retried blindly with the same brief.\n\n8. **Review what landed.** Read the diff — the hunks, not the worker's\n   transcript, from the staged index (`git diff --staged`). For anything beyond\n   an isolated change, dispatch a fresh reviewer that did not write the code,\n   a different model family per [routing.md](..\u002F..\u002Ffactory\u002Frouting.md), and\n   give it the acceptance criteria and the diff. A reviewer inheriting the\n   author's context confirms the author's assumptions.\n\n9. **Apply the lane.** [lanes.md](..\u002F..\u002Ffactory\u002Flanes.md) decides what happens\n   next: `auto` continues; `approve` posts a diff summary; `full` needs the\n   independent review from step 8. Who resolves the gate is lanes.md's call —\n   a human when attended, the agent itself under\n   [autonomy](..\u002Fplandesk-autonomy\u002FSKILL.md) with the reasoning chain posted\n   first. Either way the resolution lives as a comment, which is the human's\n   override surface.\n\n10. **Commit when the gate clears — each slice on its own.** One work item,\n    one commit, the subject naming the task; the work stays staged until its\n    gate resolves and enters history only after. Commit the moment the gate\n    clears — batching commits until the end of a run means a single later\n    failure puts every earlier success at risk, and it breaks the 1:1 between\n    history and board. Flip the task to `done` in the same step, call\n    `record_agent_progress`, and append the cycle to `.agents\u002Ffactory\u002Fruns\u002Fmetrics.jsonl`\n    (tracked — include it in the slice's commit).\n\n11. **Take the next slice, or stop.** Repeat from step 5 while the frontier has\n    work and no gate is blocking. On long runs, pulse per\n    [heartbeat.md](..\u002F..\u002Ffactory\u002Fheartbeat.md) — a worker with no output, no\n    file changes, and flat CPU is stalled rather than thinking, and the tree may\n    already hold most of its work. Close with `complete_agent_run` and report at\n    diff level — naming tasks by label, not bare id (see `.plandesk\u002Fskill.md`):\n    what shipped, what is waiting on a human, what failed and why.\n\n## Stopping\n\nStop the run and report — do not work around it — when:\n\n- a **lane gate** needs a human,\n- a worker returns **`blocked`** with a question you cannot answer from the board,\n- a gate cannot be satisfied honestly, or\n- the frontier empties.\n\n**Skip dispatch, but continue the run**, when a frontier item is `kind: 'decision'`.\nUnlike a lane gate — which blocks the tree because uncommitted work occupies it —\na decision task blocks only itself. Report it by **label** as awaiting a human\nand take the next frontier item.\n\nNever merge, and never release `scope` → `todo`; both belong to whoever owns the outcome.\n\nLeave the board true on the way out. A status that does not match what actually\nhappened is worse than no status, because the next run trusts it.\n\n## Boundaries\n\n- Groom inline, dispatch implementation. Reversing this is the common failure.\n- **Never dispatch a `kind: 'decision'` task.** Its resolution is a conversation\n  with whoever owns the outcome; a worker given one will answer its own question\n  and record a fabrication as a decision. Report it as awaiting a human and take\n  the next frontier item.\n- Do not restate the readiness bar; it lives in\n  [groom-task](..\u002Fplandesk-groom-task\u002FSKILL.md) and a second copy would drift from it.\n- Do not restate the task's spec in a brief — link it live.\n- Do not run two dispatches in one tree; give concurrent slices their own\n  worktrees.\n- Do not batch commits, and do not commit work whose gate has not cleared.\n- If a change balloons past its triaged size, return it to `scope` with the\n  reason rather than absorbing it.\n","# Foreman\n\nTakes board work to committed. Planning fills the board, the foreman runs the floor,\nworkers build, a human decides what merges.\n\nThe policy this runs on lives beside it and is not repeated here — the cycle\ncontract in [factory.md](..\u002F..\u002Ffactory\u002Ffactory.md), dispatch and verification in\n[protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md), worker selection in\n[routing.md](..\u002F..\u002Ffactory\u002Frouting.md), gates in\n[lanes.md](..\u002F..\u002Ffactory\u002Flanes.md), slice shapes in\n[slicing.md](..\u002F..\u002Ffactory\u002Fslicing.md), the brief's additions in\n[brief.md](..\u002F..\u002Ffactory\u002Fbrief.md), the worker's standard in\n[workmanship.md](..\u002F..\u002Ffactory\u002Fworkmanship.md). Read the ones a run touches.\n\n## Workflow\n\n1. **Preflight.** Confirm the tree is clean, no dispatch is already running\n   (`pgrep -f`), the board answers, and at least one worker in\n   [workers\u002F](..\u002F..\u002Ffactory\u002Fworkers\u002F) passes its probe. Each of these turns a\n   long run into wasted tokens if it fails halfway, so check before starting,\n   not on the way past. Open the run with `start_agent_run`.\n\n2. **Resolve the scope.** A named task label (looked up with `list_tasks` \u002F\n   `get_task`) works one item. `next` calls\n   `get_next_task`. `all todo` or a goal name takes the whole frontier — only\n   `todo` tasks whose prerequisites are `done`; `scope` and `backlog` are\n   waiting on a human and are not yours to release. Read each task's linked\n   spec before judging it.\n\n3. **Groom inline — do not dispatch this.** Judge each task against the\n   Definition of Ready in [groom-task](..\u002Fplandesk-groom-task\u002FSKILL.md) and rewrite\n   anything short of it with `update_task` yourself, following that skill's\n   procedure — it is the one readiness bar this project keeps, so do not invent\n   a second one here. Grooming is judgment about intent, and shipping that\n   judgment to a worker is how a plan drifts from what was actually wanted. If a\n   task cannot be groomed without a decision you do not own, return it to\n   `scope` with a comment naming the decision, and carry on with the rest.\n\n   Grooming as a prelude to dispatch does **not** wait on groom-task's own\n   `approve` gate: post the groom comment (what changed, what was inferred) for\n   traceability, then proceed — the rewrite is reviewed together with the work\n   it produced, at the item's own lane gate, where a human can still reject\n   both. A groom left uncommented is still the violation.\n\n4. **Cut slices when the frontier is wider than one item.** One task needs no\n   slicing. Several become deliverable units per\n   [slicing.md](..\u002F..\u002Ffactory\u002Fslicing.md) — complete paths through the layers\n   they touch, sized to one worker's context, grouped so parallel workers do not\n   edit the same lines. The grouping goes in each brief's WBS snapshot, not in a\n   file of its own — the task descriptions already hold the build contract, so\n   do not restate the work anywhere.\n\n5. **Dispatch implementation — default async.** One dispatch per tree, per\n   [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md). Pick the worker from\n   [routing.md](..\u002F..\u002Ffactory\u002Frouting.md) unless one was named; a named worker\n   wins, and several named workers split the slices across worktrees. Build the\n   brief from [brief-template.md](..\u002F..\u002Ffactory\u002Fbrief-template.md) — copy,\n   substitute, paste workmanship in full — which carries protocol.md's\n   five-section contract: the bar, the result contract, the ground, the WBS\n   snapshot, and a live `Context:` link.\n\n   **Fire async unless the slice is trivial.** Background is the default for\n   every real dispatch. Run foreground only when the slice is genuinely\n   small (one obvious edit, expected under a few minutes) and the user asked\n   for inline work. The recipe:\n\n   1. Reset the task's `.agents\u002Ffactory\u002Fruns\u002F` state and capture the baseline per\n      [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md) dispatch step 2 — a stale\n      result file fires the monitor instantly, and verification needs the\n      pre-dispatch test counts to diff against.\n   2. Run the worker's `probe`; substitute `{prompt_file}` and `{repo_path}` in\n      its `command` template — flags come from [workers\u002F](..\u002F..\u002Ffactory\u002Fworkers\u002F),\n      never from memory.\n   3. Append the log redirect the engine owns:\n      `> .agents\u002Ffactory\u002Fruns\u002Fworker-\u003Ctask>.log 2>&1` (never put redirects in worker files).\n   4. **Background through the harness** — `run_in_background: true` on the\n      Shell\u002FBash tool. Never append `&` or wrap in `nohup … &`; that\n      orphan-detaches, the harness fires a false \"completed\", and the real leaf\n      keeps writing to a tree nobody is watching.\n   5. **Arm the monitor in the same step**, also backgrounded — the watch loop\n      in protocol.md's *Watching a live dispatch*. Do not dispatch and \"remember\n      to watch later.\"\n   6. **Do not poll the harness completion notification.** Wrapper exit is\n      unreliable (orphan shell, transient API blip). The monitor watches\n      `.agents\u002Ffactory\u002Fruns\u002Fresult-\u003Ctask>.json`; use `Await` on the monitor shell when you are\n      ready to verify, or continue other foreman work while it runs.\n\n   **Stdin is per worker** — getting this wrong backgrounds a hang:\n\n   | delivery | workers | rule |\n   | --- | --- | --- |\n   | prompt via stdin | claude, cursor, opencode | `\u003C {prompt_file}` — **no** `\u003C \u002Fdev\u002Fnull` |\n   | prompt via `@file` or arg | pi, codex, grok | pi: `@{prompt_file}`; codex: `\"$(cat …)\"`; grok: `--prompt-file` — add `\u003C \u002Fdev\u002Fnull` |\n\n   Mint the live link explicitly: `create_share_link` on the task (or the goal,\n   for a multi-slice run), `expires: 7d`, and put the returned `markdown_url`\n   in the brief as `Context:`. The worker has no MCP access, so this is the\n   only way it reads live board state instead of a copy that goes stale the\n   moment someone edits the task. Derive the WBS snapshot from `list_tasks` +\n   `list_edges` so the worker sees the agreed order and the paths a later item\n   owns; a worker given one node and no map finishes the next one too.\n\n   **Never dispatch a `kind: 'decision'` task.** Its resolution is a\n   conversation with whoever owns the outcome; a worker given one will answer\n   its own question and record a fabrication as a decision. Report it by task\n   **label** as awaiting a human and take the next frontier item — a decision\n   task blocks only itself; unlike a lane gate it does not stop the run.\n\n6. **Stage the moment a worker returns, before reading anything.** Review takes\n   minutes and unstaged work is defenceless for all of them; staged work\n   survives a later `git checkout` because git restores it from the index. This\n   ordering is the cheapest real protection in the whole run.\n\n   **Unless the dispatch was killed.** No result file means it ran no gates and\n   wrote no result, so the tree holds unverified partial output rather than work\n   product. Discard it and re-dispatch clean per\n   [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md) — staging it makes salvage look like\n   a deliverable, and the next reader cannot tell the difference.\n\n7. **Verify the claims.** Follow the verification sequence in\n   [protocol.md](..\u002F..\u002Ffactory\u002Fprotocol.md) — gate integrity before re-running\n   anything, then suppressions, then per-package test counts against the\n   pre-dispatch baseline, then debris. Exit codes decide, not the worker's\n   summary. A dispatch that fails verification is recorded and re-scoped, not\n   retried blindly with the same brief.\n\n8. **Review what landed.** Read the diff — the hunks, not the worker's\n   transcript, from the staged index (`git diff --staged`). For anything beyond\n   an isolated change, dispatch a fresh reviewer that did not write the code,\n   a different model family per [routing.md](..\u002F..\u002Ffactory\u002Frouting.md), and\n   give it the acceptance criteria and the diff. A reviewer inheriting the\n   author's context confirms the author's assumptions.\n\n9. **Apply the lane.** [lanes.md](..\u002F..\u002Ffactory\u002Flanes.md) decides what happens\n   next: `auto` continues; `approve` posts a diff summary; `full` needs the\n   independent review from step 8. Who resolves the gate is lanes.md's call —\n   a human when attended, the agent itself under\n   [autonomy](..\u002Fplandesk-autonomy\u002FSKILL.md) with the reasoning chain posted\n   first. Either way the resolution lives as a comment, which is the human's\n   override surface.\n\n10. **Commit when the gate clears — each slice on its own.** One work item,\n    one commit, the subject naming the task; the work stays staged until its\n    gate resolves and enters history only after. Commit the moment the gate\n    clears — batching commits until the end of a run means a single later\n    failure puts every earlier success at risk, and it breaks the 1:1 between\n    history and board. Flip the task to `done` in the same step, call\n    `record_agent_progress`, and append the cycle to `.agents\u002Ffactory\u002Fruns\u002Fmetrics.jsonl`\n    (tracked — include it in the slice's commit).\n\n11. **Take the next slice, or stop.** Repeat from step 5 while the frontier has\n    work and no gate is blocking. On long runs, pulse per\n    [heartbeat.md](..\u002F..\u002Ffactory\u002Fheartbeat.md) — a worker with no output, no\n    file changes, and flat CPU is stalled rather than thinking, and the tree may\n    already hold most of its work. Close with `complete_agent_run` and report at\n    diff level — naming tasks by label, not bare id (see `.plandesk\u002Fskill.md`):\n    what shipped, what is waiting on a human, what failed and why.\n\n## Stopping\n\nStop the run and report — do not work around it — when:\n\n- a **lane gate** needs a human,\n- a worker returns **`blocked`** with a question you cannot answer from the board,\n- a gate cannot be satisfied honestly, or\n- the frontier empties.\n\n**Skip dispatch, but continue the run**, when a frontier item is `kind: 'decision'`.\nUnlike a lane gate — which blocks the tree because uncommitted work occupies it —\na decision task blocks only itself. Report it by **label** as awaiting a human\nand take the next frontier item.\n\nNever merge, and never release `scope` → `todo`; both belong to whoever owns the outcome.\n\nLeave the board true on the way out. A status that does not match what actually\nhappened is worse than no status, because the next run trusts it.\n\n## Boundaries\n\n- Groom inline, dispatch implementation. Reversing this is the common failure.\n- **Never dispatch a `kind: 'decision'` task.** Its resolution is a conversation\n  with whoever owns the outcome; a worker given one will answer its own question\n  and record a fabrication as a decision. Report it as awaiting a human and take\n  the next frontier item.\n- Do not restate the readiness bar; it lives in\n  [groom-task](..\u002Fplandesk-groom-task\u002FSKILL.md) and a second copy would drift from it.\n- Do not restate the task's spec in a brief — link it live.\n- Do not run two dispatches in one tree; give concurrent slices their own\n  worktrees.\n- Do not batch commits, and do not commit work whose gate has not cleared.\n- If a change balloons past its triaged size, return it to `scope` with the\n  reason rather than absorbing it.\n",{"name":5,"description":7,"argument-hint":24,"user-invocable":25},"\u003Ctask id | 'next' | 'all todo' | goal name> [--to \u003Cworker>]",true,"main","1e2d885d184029608df0aab985a4a1545c3a1a1a",".agents\u002Fskills\u002Fplandesk-foreman",1,"https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk","https:\u002F\u002Fplandesk-docs.pages.dev","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering","2026-10-04 12:26:19",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":25},[36,39,42],{"tool":37,"command":38},"npx","npx skills add asyncdotengineering\u002Fplandesk --skill plandesk-foreman",{"tool":40,"command":41},"bunx","bunx skills add asyncdotengineering\u002Fplandesk --skill plandesk-foreman",{"tool":43,"command":44},"pnpm","pnpm dlx skills add asyncdotengineering\u002Fplandesk --skill plandesk-foreman","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fplandesk-foreman\u002FSKILL.md",[],[48,55,61,67,73,79],{"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},"01M75CB59TR0Q4BQ056JTQKRB3","connecting-agents","Agent conventions for Plan Desk MCP — task labels, docs, notes, files, artifacts, sharing, edges, and agent runs.","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002Fapps\u002Fdocs\u002Fsrc\u002Fcontent\u002Fdocs\u002Fconnecting-agents\u002Fskill.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M75CB3ZV1RWR3SJPG2PMBJRG",".plandesk","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.plandesk\u002Fskill.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":64,"categories":65,"keywords":66,"publishedAt":19},"01M75CB3428SD8HWHTBW857YXA","plandesk","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fplandesk\u002FSKILL.md",[],[],{"skillNo":68,"name":69,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":19},"01M75CB277DZXMNDYF9FNDRN84","plandesk-timebox","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fplandesk-timebox\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M75CB03K0H1H0Y5F4WR9TFZZ","plandesk-standup","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fplandesk-standup\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":6,"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},"01M75CAXTZWTY17EY7T4SNA55A","plandesk-standdown","https:\u002F\u002Fgithub.com\u002Fasyncdotengineering\u002Fplandesk\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fplandesk-standdown\u002FSKILL.md",[],[]]