@sakanano-moe/agenthelm 6.7.0 → 6.11.0-rc.1

package/runtime/plugins/project-orchestrator/sop-templates/blank/SOP.md

@@ -0,0 +1,57 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "{{title}}"
+status: "draft"
+version: "1.0.0"
+description: "【待补充】一句话：这个流程做什么、什么时候该想起它"
+trigger: "【待补充】什么情况下用这个流程"
+exclude: ""
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases: []
+examples_hit: []
+examples_miss: []
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## 目的
+
+【待补充】这个流程要达成什么结果，一两句话。
+
+## 前置条件
+
+每行一项：`- 种类：名称｜用途：…｜安装：…｜替代：另一项的名称｜缺失时：询问用户`。
+种类可用：命令、模块（Python 模块）、环境变量、路径、技能、MCP、手动确认。缺失时可用：询问用户、停止、跳过该步、使用替代。
+API 密钥只写环境变量名，不写密钥本身。确实不需要任何工具时写“无”。
+
+- 命令：【待补充】｜用途：【待补充】｜安装：【待补充】｜缺失时：询问用户
+
+## 步骤
+
+每步一个编号，下面用缩进的子项补充；可用子项：工具、输入、中间产物、最终产物、检查点、失败时、需确认、说明。
+“工具”里写的名称必须在上面的前置条件里声明过。
+
+1. 【待补充】第一步做什么
+   - 工具：【待补充】
+   - 输入：【待补充】
+   - 中间产物：【待补充】
+   - 检查点：【待补充】怎样算这一步做对了
+   - 失败时：【待补充】（不写则按“执行约定”处理）
+
+## 完成标准
+
+- 【待补充】怎样才算整个流程做完了（能被检查的条件，例如“成品文件存在于 …”）
+
+## 例外与回退
+
+- 【待补充】常见的例外情况和处理办法；没有就写“无”。
+
+## 修订记录
+
+暂无。

package/runtime/plugins/project-orchestrator/sop-templates/knowledge-digest/SOP.en.md

@@ -0,0 +1,77 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "Weekly knowledge digest"
+status: "draft"
+version: "1.0.0"
+description: "Regularly tidy new notes and materials: merge duplicates, fill in links and distill rule candidates"
+trigger: "The user wants the recent notes and materials tidied, or the fixed digest day has come"
+exclude: "Only looking up one particular document"
+schedule: "weekly:sat"
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases:
+  - "knowledge digest"
+  - "tidy notes"
+examples_hit:
+  - "Tidy up my recent notes"
+  - "Do a knowledge digest"
+examples_miss:
+  - "Find that report from last time"
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## Purpose
+
+Bring the notes and materials added or changed over a period into a state where they are searchable, free of duplicates and linked to each other, and distill practices that keep recurring into rule candidates.
+
+## Prerequisites
+
+- Command: agenthelm | Purpose: find new materials by keyword and date (`agenthelm project find`) | Install: see agenthelm's INSTALL.md | If missing: Ask the user
+
+## Steps
+
+1. Settle the scope of the digest
+   - Inputs: a directory, tags or a period
+   - Checkpoint: the user states the scope explicitly
+   - Needs confirmation: yes
+   - Note: if the scope contains materials with personal progress, accounts or private material, tell the user first and let them decide whether to include them; by default this procedure tidies only shareable knowledge content
+2. Collect new and changed materials
+   - Tools: agenthelm
+   - Inputs: the scope; the date of the last digest
+   - Intermediates: a list of things to tidy
+   - Checkpoint: the list holds only materials in scope
+3. Merge duplicates and outdated content
+   - Inputs: the list of things to tidy
+   - Checkpoint: every merge or deletion is on the list
+   - Needs confirmation: yes
+   - Note: hand the list to the user for confirmation before merging or deleting, and never delete directly; keep the originals that were merged until the user confirms
+4. Fill in links and indexes
+   - Checkpoint: each new material is referenced by at least one index or related material, and no link is broken
+5. Distill rule candidates
+   - Inputs: practices that recurred during the digest and that the user has stated explicitly
+   - Intermediates: a candidate list (each with the places it is based on)
+   - Note: only form candidates and hand them to rule management (opinion-manager) for the user-confirmation flow; never write into OPINION.md yourself, and the number of recurrences is not the user's confirmation
+6. Record the result
+   - Final outputs: docs/digest-DATE.md inside this Case's directory (using the structure of templates/digest.md in this directory)
+   - Checkpoint: merges, deletions, new links and candidates are each recorded
+
+## Completion criteria
+
+- The result is saved and registered as a deliverable
+- Everything merged or deleted was confirmed by the user
+- The rule candidates were handed to the user for a decision, or "no candidates this time" is recorded explicitly
+
+## Exceptions and fallback
+
+- Nothing new in scope: record it as it is, and do not change existing content just to have something to show
+- Content that was already tidied turns out to be wrong: do not overwrite the original; point it out in the result and ask the user
+
+## Revision log
+
+None yet.

package/runtime/plugins/project-orchestrator/sop-templates/knowledge-digest/SOP.md

@@ -0,0 +1,77 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "每周知识整理"
+status: "draft"
+version: "1.0.0"
+description: "定期整理新增的笔记与资料：合并重复、补齐链接，并提炼规则候选"
+trigger: "用户要整理最近的笔记和资料，或到了固定的整理日"
+exclude: "只是查找某一份资料"
+schedule: "weekly:sat"
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases:
+  - "知识整理"
+  - "整理笔记"
+examples_hit:
+  - "把最近的笔记整理一下"
+  - "做一次知识整理"
+examples_miss:
+  - "帮我找一下上次那份报告"
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## 目的
+
+把一段时间内新增或变更的笔记、资料整理成可检索、不重复、互相链接的状态，并把反复出现的做法沉淀成规则候选。
+
+## 前置条件
+
+- 命令：agenthelm｜用途：按关键词、日期查找新增的资料（`agenthelm project find`）｜安装：见 agenthelm 的 INSTALL.md｜缺失时：询问用户
+
+## 步骤
+
+1. 确定整理范围
+   - 输入：目录、标签或时间段
+   - 检查点：范围由用户明确给出
+   - 需确认：是
+   - 说明：范围内如有含个人进度、账号或私人素材的资料，先告诉用户，由用户决定是否纳入；本流程默认只整理可共享的知识类内容
+2. 收集新增与变更的资料
+   - 工具：agenthelm
+   - 输入：范围；上次整理的日期
+   - 中间产物：待整理清单
+   - 检查点：清单只含范围内的资料
+3. 合并重复与过时内容
+   - 输入：待整理清单
+   - 检查点：每一处合并或删除都列在清单里
+   - 需确认：是
+   - 说明：合并或删除前把清单交给用户确认，不直接删除；被合并的原文先保留到用户确认之后
+4. 补齐链接与索引
+   - 检查点：新增资料至少被一个索引或相关资料引用，没有断链
+5. 提炼规则候选
+   - 输入：整理过程中反复出现、用户明确表达过的做法
+   - 中间产物：候选清单（每条写明依据来自哪几处）
+   - 说明：只形成候选，交给规则管理（opinion-manager）走用户确认流程；不自行写入 OPINION.md，重复出现的次数不等于用户已确认
+6. 记录整理结果
+   - 最终产物：本 Case 目录内的 docs/digest-日期.md（使用本目录 templates/digest.md 的结构）
+   - 检查点：合并、删除、新增链接、候选各有记录
+
+## 完成标准
+
+- 整理结果已保存并登记为交付物
+- 合并或删除的内容都经用户确认
+- 规则候选已交给用户裁决，或明确记录“本次无候选”
+
+## 例外与回退
+
+- 范围内没有新增内容：照实记录，不为了凑数改动已有内容
+- 发现已整理的内容有误：不覆盖原文，在结果里标出并问用户
+
+## 修订记录
+
+暂无。

package/runtime/plugins/project-orchestrator/sop-templates/knowledge-digest/templates/digest.en.md

@@ -0,0 +1,11 @@
+# Knowledge digest (date range)
+
+## Scope
+
+## Merges and deletions (all confirmed)
+
+## New links and indexes
+
+## Rule candidates (awaiting the user's confirmation)
+
+## Open questions

package/runtime/plugins/project-orchestrator/sop-templates/knowledge-digest/templates/digest.md

@@ -0,0 +1,11 @@
+# 知识整理（日期范围）
+
+## 范围
+
+## 合并与删除（均经确认）
+
+## 新增链接与索引
+
+## 规则候选（待用户确认）
+
+## 遗留问题

package/runtime/plugins/project-orchestrator/sop-templates/video-to-transcript/SOP.en.md

@@ -0,0 +1,87 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "Video to transcript"
+status: "draft"
+version: "1.0.0"
+description: "Turn a video (a link or a local file) into a transcript"
+trigger: "The user wants a video turned into a transcript, subtitles or searchable text"
+exclude: "The user already has a transcript and only wants it summarized or rewritten"
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases:
+  - "video to text"
+  - "video transcription"
+examples_hit:
+  - "Turn this video into a transcript"
+  - "Transcribe the video at this link"
+examples_miss:
+  - "Summarize this transcript"
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## Purpose
+
+Turn a video into a readable, searchable transcript that keeps its source information.
+
+## Prerequisites
+
+A set of common choices; the names are examples: confirm, replace or delete each one according to the tools you actually use. Items marked "Only if" are needed only when this run's input matches the condition.
+
+- Command: yutto | Purpose: download Bilibili videos (BV / av IDs, bilibili.com links); `--audio-only` fetches just the audio, `-d` sets the output directory | Install: `pipx install yutto` (needs FFmpeg configured) | Only if: the input is a Bilibili video | If missing: Ask the user
+- Command: yt-dlp | Purpose: download online videos from other sites | Install: `pipx install yt-dlp` or the system package manager | Only if: the input is an online video from a site other than Bilibili | If missing: Ask the user
+- Command: ffmpeg | Purpose: merge Bilibili audio and video streams, extract audio from a video, convert to the format the transcription tool needs | Install: the system package manager (for example `brew install ffmpeg`) | If missing: Ask the user
+- Command: whisper-cli | Purpose: transcription with a local model | Install: follow the official instructions of the local transcription tool you use | Alternative: TRANSCRIBE_API_KEY | If missing: Use the alternative
+- Path: ~/models/whisper.bin | Purpose: the model file whisper-cli needs; change it to your actual location | If missing: Ask the user
+- Environment variable: TRANSCRIBE_API_KEY | Purpose: the key for a cloud transcription API, if you switch to one (write only the variable name) | If missing: Ask the user
+
+## Steps
+
+1. Confirm the input and options
+   - Inputs: the video link or local file path; the language; whether timestamps are needed
+   - Checkpoint: the input is clear, and a local file really exists
+2. Get the video file
+   - Tools: yutto, yt-dlp
+   - Inputs: the video link (skip this step for a local file)
+   - Intermediates: the downloaded video or audio file
+   - Checkpoint: the file exists and is larger than 0 bytes
+   - On failure: Ask the user (when a link is dead, needs a login or is region-locked, do not work around it yourself)
+   - Needs confirmation: yes
+   - Note: use yutto for Bilibili links and yt-dlp for other sites. For content that needs a login, the user logs in and saves the login information themselves (see yutto's official documentation for how); credentials are written neither into this SOP nor into the Case. Downloading a large file is a slow step, so the "run conventions" decide whether to ask first
+3. Extract the audio
+   - Tools: ffmpeg
+   - Inputs: the video or audio file from the previous step
+   - Intermediates: audio.wav (mono, 16 kHz; follow the transcription tool if it needs another format)
+   - Checkpoint: the audio length differs from the video length by at most 1 second
+4. Transcribe
+   - Tools: whisper-cli, TRANSCRIBE_API_KEY
+   - Inputs: audio.wav; the language
+   - Intermediates: the raw transcription result
+   - Checkpoint: the text is not empty and its length roughly matches the audio length
+   - Note: prefer local transcription; switch to a cloud API only when the local one is unavailable and the user agrees, because using an API sends the audio to a third-party service, which needs the user's consent first
+5. Tidy up and save
+   - Inputs: the raw transcription result
+   - Final outputs: the transcript (see the completion criteria for where to save it)
+   - Checkpoint: spot-check one passage each from the start, the middle and the end against the audio
+6. Clean up
+   - Note: handle the intermediates according to the "run conventions"; clean up only files registered in the Case during this run
+
+## Completion criteria
+
+- The transcript is saved at: [TO FILL IN]
+- The start of the transcript records the source (link or file name), the transcription tool and the date
+- The intermediates were handled as agreed, and the outcome is noted in the Case's work log
+
+## Exceptions and fallback
+
+- The video has no audio track: stop and tell the user; do not make up text
+- The audio is very long: transcribe it in segments and merge them, then check that the passages join up
+
+## Revision log
+
+None yet.

package/runtime/plugins/project-orchestrator/sop-templates/video-to-transcript/SOP.md

@@ -0,0 +1,87 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "视频转文字稿"
+status: "draft"
+version: "1.0.0"
+description: "把一个视频（链接或本地文件）转成文字稿"
+trigger: "用户要把某个视频转成文字稿、字幕或可检索的文本"
+exclude: "用户已经提供了文字稿，只是要总结或改写"
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases:
+  - "视频转文字"
+  - "视频转录"
+examples_hit:
+  - "帮我把这个视频转成文字稿"
+  - "这个链接的视频转录一下"
+examples_miss:
+  - "总结一下这份文字稿"
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## 目的
+
+把一个视频转成可阅读、可检索的文字稿，并保留来源信息。
+
+## 前置条件
+
+下面是一组常见选择，名称都是示例：请按你实际使用的工具逐项确认、替换或删除。带“仅当”的项只在本次输入符合条件时才需要。
+
+- 命令：yutto｜用途：下载 B 站视频（BV / av 号、bilibili.com 链接），可用 `--audio-only` 只取音频、`-d` 指定输出目录｜安装：`pipx install yutto`（需要已配置 FFmpeg）｜仅当：输入是 B 站视频｜缺失时：询问用户
+- 命令：yt-dlp｜用途：下载其他网站的在线视频｜安装：`pipx install yt-dlp` 或系统包管理器｜仅当：输入是 B 站以外的在线视频｜缺失时：询问用户
+- 命令：ffmpeg｜用途：合并 B 站音视频流、从视频提取音频、转成转录工具需要的格式｜安装：系统包管理器（如 `brew install ffmpeg`）｜缺失时：询问用户
+- 命令：whisper-cli｜用途：本地模型转录｜安装：按所用的本地转录工具的官方说明｜替代：TRANSCRIBE_API_KEY｜缺失时：使用替代
+- 路径：~/models/whisper.bin｜用途：whisper-cli 需要的模型文件，请改成你的实际位置｜缺失时：询问用户
+- 环境变量：TRANSCRIBE_API_KEY｜用途：改用云端转录 API 时的密钥（只写变量名）｜缺失时：询问用户
+
+## 步骤
+
+1. 确认输入与选项
+   - 输入：视频链接或本地文件路径；语言；是否需要时间戳
+   - 检查点：输入明确，本地文件确实存在
+2. 取得视频文件
+   - 工具：yutto、yt-dlp
+   - 输入：视频链接（本地文件则跳过本步）
+   - 中间产物：下载的视频或音频文件
+   - 检查点：文件存在且大小大于 0
+   - 失败时：询问用户（链接失效、需要登录或地区限制时不要自行绕过）
+   - 需确认：是
+   - 说明：B 站链接用 yutto，其他站点用 yt-dlp。需要登录才能看的内容，由用户自己登录并保存登录信息（yutto 的做法见其官方文档），登录凭据不写进本 SOP、不写进 Case。大文件下载属于耗时步骤，按“执行约定”决定是否先问
+3. 提取音频
+   - 工具：ffmpeg
+   - 输入：上一步的视频或音频文件
+   - 中间产物：audio.wav（单声道，16 kHz；转录工具要求其他格式时按其要求）
+   - 检查点：音频时长与视频时长相差不超过 1 秒
+4. 转录
+   - 工具：whisper-cli、TRANSCRIBE_API_KEY
+   - 输入：audio.wav；语言
+   - 中间产物：原始转录结果
+   - 检查点：文本非空，长度与音频时长大致匹配
+   - 说明：优先使用本地转录；本地不可用且用户同意时才改用云端 API，使用 API 会把音频发送给第三方服务，需要先问用户
+5. 整理与保存
+   - 输入：原始转录结果
+   - 最终产物：文字稿（保存位置见完成标准）
+   - 检查点：抽查开头、中间、结尾各一段与音频一致
+6. 清理
+   - 说明：按“执行约定”处理中间产物；只清理本次在 Case 里登记过的文件
+
+## 完成标准
+
+- 文字稿已保存到：【待补充】
+- 文字稿开头记录了来源（链接或文件名）、转录工具与日期
+- 中间产物已按约定处理，处理结果记在 Case 的工作记录里
+
+## 例外与回退
+
+- 视频无音轨：停止并告诉用户，不要凭空生成文本
+- 音频很长：可按段切分转录再合并，合并后检查段落衔接
+
+## 修订记录
+
+暂无。

package/runtime/plugins/project-orchestrator/sop-templates/weekly-review/SOP.en.md

@@ -0,0 +1,75 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "Weekly review"
+status: "draft"
+version: "1.0.0"
+description: "Look back on the progress of chosen items each week, write a review and plan the next week"
+trigger: "The user wants a weekly review or to look back on the week, or the fixed review day has come"
+exclude: "Only asking where one item stands right now"
+schedule: "weekly:sun"
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases:
+  - "weekly review"
+examples_hit:
+  - "Do this week's review"
+  - "Look back on what I did this week"
+examples_miss:
+  - "Where is this item right now"
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## Purpose
+
+Look back on what the chosen items actually achieved this week and where they got stuck, collect the lessons, and settle the few most important things for next week.
+
+## Prerequisites
+
+- Command: agenthelm | Purpose: read the progress of the chosen items (`agenthelm project progress --output json`) | Install: see agenthelm's INSTALL.md | If missing: Ask the user
+
+## Steps
+
+1. Settle the scope of the review
+   - Inputs: the items to include (tags, item names or IDs)
+   - Checkpoint: the user states the scope explicitly; do not include every item by default
+   - Needs confirmation: yes
+   - Note: the review reads the work logs of the chosen items, which may hold personal progress. Items the user did not include are never read and never written into the review; the review document stays in the project that runs it, and if that project may become public, do not use this procedure to review personal progress
+2. Collect the progress of the chosen items
+   - Tools: agenthelm
+   - Inputs: this week's start and end dates; the scope from the previous step
+   - Intermediates: a progress list (item, status, this week's work log)
+   - Checkpoint: only items in scope are covered, and all of them that are in progress or were closed this week
+3. Compare with last week's plan
+   - Inputs: the "next week's plan" in the previous review document
+   - Checkpoint: every planned item has a verdict: done / not done, with the reason
+4. Write the review document
+   - Inputs: the results of the steps above
+   - Final outputs: docs/review-DATE.md inside this Case's directory (using the structure of templates/review.md in this directory)
+   - Checkpoint: every section has content and the conclusions have grounds (pointing to an item or a file); the scope of this review is stated at the top
+5. Plan next week
+   - Inputs: the unfinished items and blockers in the review document
+   - Checkpoint: no more than the number the user set; every item has a clear next action
+   - Needs confirmation: yes
+   - Note: the user decides next week's plan; only suggest
+6. Register the deliverable
+   - Note: register the review document as a deliverable of the Case
+
+## Completion criteria
+
+- The review document is saved inside this Case's directory, states its scope at the top and is registered as a deliverable
+- Next week's plan was confirmed by the user
+
+## Exceptions and fallback
+
+- The chosen items have no records at all this week: say "no records this week" as it is, and ask the user whether something went unrecorded
+- The user later wants to publish or share the review: the user first confirms, section by section, what may be public, and then it is copied out; the original document is not moved
+
+## Revision log
+
+None yet.

package/runtime/plugins/project-orchestrator/sop-templates/weekly-review/SOP.md

@@ -0,0 +1,75 @@
+---
+schema_version: "1.0"
+id: "{{id}}"
+title: "每周复盘"
+status: "draft"
+version: "1.0.0"
+description: "每周回看所选事项的进展，写复盘并规划下周"
+trigger: "用户要做周复盘、回顾本周，或到了固定的复盘日"
+exclude: "只是查询某个事项当前进展"
+schedule: "weekly:sun"
+confirm: "ask"
+creates_case: "always"
+created: "{{date}}"
+aliases:
+  - "周复盘"
+examples_hit:
+  - "做一下本周复盘"
+  - "回顾一下这周做了什么"
+examples_miss:
+  - "这个事项现在到哪了"
+on_events: []
+outputs: []
+sources: []
+decided: []
+---
+# {{title}}
+
+## 目的
+
+回看本周所选事项实际完成了什么、卡在哪里，沉淀经验，并定下下周最重要的几件事。
+
+## 前置条件
+
+- 命令：agenthelm｜用途：读取所选事项的进展（`agenthelm project progress --output json`）｜安装：见 agenthelm 的 INSTALL.md｜缺失时：询问用户
+
+## 步骤
+
+1. 确定复盘范围
+   - 输入：要纳入的事项（标签、事项名或编号）
+   - 检查点：范围由用户明确给出，不默认纳入全部事项
+   - 需确认：是
+   - 说明：复盘会读取所选事项的工作记录，其中可能有个人进度。未被用户纳入的事项一律不读、不写进复盘；复盘文档只留在运行它的这个项目里，该项目若会公开，不要用本流程复盘个人进度
+2. 收集所选事项的进展
+   - 工具：agenthelm
+   - 输入：本周的起止日期；上一步确定的范围
+   - 中间产物：进展清单（事项、状态、本周的工作记录）
+   - 检查点：只覆盖范围内的事项，且覆盖其中所有进行中和本周关闭的
+3. 对照上周计划
+   - 输入：上一次复盘文档里的“下周计划”
+   - 检查点：每条计划都有结论：完成 / 未完成及原因
+4. 写复盘文档
+   - 输入：以上步骤的结果
+   - 最终产物：本 Case 目录内的 docs/review-日期.md（使用本目录 templates/review.md 的结构）
+   - 检查点：所有小节都有内容，结论有依据（指向事项或文件）；开头写明本次范围
+5. 规划下周
+   - 输入：复盘文档里的未完成项与阻塞
+   - 检查点：不超过用户设定的数量；每项有明确的下一动作
+   - 需确认：是
+   - 说明：下周计划由用户决定，只提出建议
+6. 登记交付物
+   - 说明：把复盘文档登记为 Case 的交付物
+
+## 完成标准
+
+- 复盘文档已保存在本 Case 目录内，开头写明了范围，并登记为交付物
+- 下周计划已经用户确认
+
+## 例外与回退
+
+- 所选事项本周没有任何记录：照实写“本周无记录”，并提示用户是否漏记
+- 用户后来想把复盘公开或分享：先由用户逐段确认哪些内容可以公开，再复制出去，不直接移动原文档
+
+## 修订记录
+
+暂无。

package/runtime/plugins/project-orchestrator/sop-templates/weekly-review/templates/review.en.md

@@ -0,0 +1,11 @@
+# Weekly review (date range)
+
+## Done this week
+
+## Not done, and why
+
+## Blockers and help needed
+
+## Lessons and improvements
+
+## Next week's plan

package/runtime/plugins/project-orchestrator/sop-templates/weekly-review/templates/review.md

@@ -0,0 +1,12 @@
+# 周复盘 （日期范围）
+
+## 本周完成
+
+## 未完成与原因
+
+## 阻塞与需要的帮助
+
+## 经验与改进
+
+## 下周计划
+

package/runtime/plugins/project-orchestrator/templates/CASE.en.md

@@ -0,0 +1,39 @@
+# Summary
+
+{{title}}; current state: {{current_state}}; next: {{next_action}}.
+
+## Background and scope
+
+Source: {{source}}. Goal: {{goal}}. Definition of done: {{done_when}}.
+
+## Current trusted state
+
+{{current_state}}
+
+## Next action
+
+{{next_action}}
+
+## Blockers and resume conditions
+
+None.
+
+## Decision log
+
+None yet.
+
+## Work log
+
+- {{created_at}} Case created.
+
+## Deliverables and quality gates
+
+Pending.
+
+## Result
+
+The definition of done is not met yet.
+
+## Lessons candidates
+
+None yet.