task-handoff
Skill0 starsChanged 4 months agoMITView on GitHub
When the user finishes planning and says to hand a task off to whatever executor they name — a separate model/agent/tool that shares the project directory ("把这个交给 qoder 干", "丢给千问做", "交给 codex", "让 gemini 去实现", "dispatch this to X") — this skill extracts that executor name dynamically (the names here are only examples, never a fixed whitelist), auto-creates a task folder named after it, writes a scoped PROMPT.md constraining what that executor may do, and outputs a ready-to-paste launch prompt the user forwards to it. Also handles later batch review by executor ("验收昨天交给 qoder 的任务", "千问那些任务咋样了", "codex 做完没") — scanning that executor's recent task folders and reading each DONE.md. The two agents never talk directly. Not for compacting the current conversation into a session handoff.
6 other projects use a version of this.
Adopted by
Most recently taken up by
- PaulRBerg/dot-agentstheir own version, first seen Nov 6, 2026
- PaulRBerg/agent-skillstheir own version, first seen Oct 24, 2026
- aberson/skill-meshtheir own version, first seen Oct 24, 2026
- 58254727-blip/codex-task-control-towertheir own version, first seen Oct 13, 2026
- Peter537/peter537-agent-plugintheir own version, first seen Oct 11, 2026
- yearth/skillstheir own version, first seen Sep 8, 2026
Other versions6 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 6
- PaulRBerg/agent-skills★ 104same name
- PaulRBerg/dot-agents★ 6same name
- yearth/skills★ 4same name
- aberson/skill-mesh★ 2same name
- 58254727-blip/codex-task-control-tower★ 1same name
- Peter537/peter537-agent-plugin★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
Files
Task Handoff
规划由你(本 Agent)完成,执行交给用户当下点名的那个执行方(qoder、千问、codex、gemini… 都行——名字由用户那句话决定,本 skill 不预设也不限定清单)。两边不对话,靠共享目录里的文件交接。你负责:下发时建好文件+吐转交 prompt,验收时按执行者批量看结果。
下发:用户说「把这个丢给 X 干」
一句话触发。X 是用户点名的执行方(qoder、千问、codex、gemini… 仅为举例,任何名字都行——照用户说的提取,不比对清单)。自动做三件事:
- 建文件夹(不存在则新建):
<project-root>/tasks-for-<executor>-night-<N>/task-<seq>_<slug>/<project-root>:运行时解析当前项目根,不要写死任何固定路径。取当前工作目录所在的 git 仓库根(git rev-parse --show-toplevel);不在 git 仓库内则退回当前工作目录。执行侧与规划侧共享的就是这个根。<executor>:从用户那句话里提取执行方名字,转成 kebab-case slug("qoder"→qoder,"千问"→qwen,"codex"→codex)。这些只是示例,不是白名单——用户说谁就用谁。<N>:夜次编号。扫描现有tasks-for-<executor>-night-*,同一次派活(同一晚)复用当前最大编号;开新一晚则最大编号 +1(没有任何则从 1 开始)。<seq>:任务序号。扫描所有已存在的task-*(跨全部夜次),取最大 +1,全局递增。<slug>:简短 kebab-case 任务名
- 写
PROMPT.md到该文件夹(模板见 assets/templates/PROMPT.md),约束执行侧:- frontmatter:
task/executor/dispatched(日期)/type(deterministic或exploratory)/scope(允许改动的路径)/status: pending - 正文:目标 / 完成定义 / 范围约束 / 背景(摘录进正文,不引用外部文档——执行侧读不到)/ 验证方式 / 产出要求(写回 DONE.md)
- 硬约束:没有「完成定义+验证方式」不下发;
scope外的路径禁止动。
- frontmatter:
- 吐一段转交 prompt 直接输出在对话里,让用户复制粘贴给执行模型。格式:
用绝对路径,执行模型才能直接定位。你有一个新任务。请先读取并严格遵守: <绝对路径>/tasks-for-<executor>-night-<N>/task-<seq>_<slug>/PROMPT.md 只允许改动 PROMPT.md 中 scope 列出的路径,禁止越界。 完成后在该任务文件夹下新建 <executor>-<YYYYMMDD-HHMM>/ 子目录,在里面写 DONE.md (含 status: success|partial|failed、改了什么、怎么验证、结论)。
若用户说的执行方是本环境能直接调用的工具(如 codex skill),可以直接调它执行,而不必让用户手动转交——按当前可用工具判断。默认(千问等无接口的模型)走"吐转交 prompt 让用户手动丢过去"。
验收:用户说「验收昨天交给千问的任务」
或任何提到某个非当前执行模型在处理的事项("千问那些咋样了"、"codex 做完没")。自动:
- 定位执行者:从话里提取用户点名的执行方,转成同样的 slug(qoder→
qoder、千问→qwen…),据此定位tasks-for-<executor>-night-*。 - 扫目录:列出
tasks-for-<executor>-night-*/task-*/下的任务("昨天/昨晚"→对应夜次;没说就看最近几晚里PROMPT.md的status: pending、还没验收的)。 - 逐个读:每个任务读执行侧产出子目录里的
DONE.md(即<executor>-<时间戳>/DONE.md)。- 有
DONE.md→ 看status,对照该任务的PROMPT.md的「验证方式」核对「怎么验证」是否真做了、结果对得上。 - 没
DONE.md→ 标记为"未回/未完成"。
- 有
- 汇总回报:一次性给出——哪些
success、哪些partial/failed(卡在哪、要不要你拍板)、哪些没回。逐条不用你再点名。 - 不要只看
status: success就放行——验证描述含糊或和 PROMPT 的验证方式对不上,按partial处理并说明。
partial/failed 且需要重做时:在同文件夹新写一版 PROMPT.md(只写这次要修正什么,不照抄原版),再吐一段转交 prompt。
模板
- assets/templates/PROMPT.md — 下发约束文件
- assets/templates/DONE.md — 执行侧回写文件(放模板供参考,执行侧照此产出)
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

