Todolist Md Clawdbot Copy — Todo列出 Md Clawd机器人 Copy

todo列出-md-clawd机器人

Operate on todo列出-md: a Markdown-first todo viewer/editor. The 应用 does not contAIn an AI; the 机器人 reads/writes Markdown.

Operating rules (must follow) Markdown is the 系统 of record. It’s fine to discuss in chat, but final answers/decisions must be written into Markdown. Use only 机器人 markers of this form. Do not introduce other syntaxes (no Question[id=...], no custom metadata blocks). Preserve task 身份. todo列出-md derives task IDs from Markdown line positions. Some storage backends also have a stable 身份 key (e.g. Drive fileId, local path, S3 bucket+key). Do not “replace” a file in a way that changes its 身份 key unless the user explicitly accepts it. Keep write-back edits line-stable. Avoid 添加ing/removing lines inside an existing task item or its description blockquote. Prefer single-line, in-place edits (edit text on an existing line). Last review stamp (Option B: top-of-file header line) Goal: a single line near the top recording last 机器人 review time. Rule: never insert a new line once the header exists. Only 更新 the existing header line. If the header does not exist, you may insert it at the very top only if the user explicitly opted into Option B. Canonical 格式化: 2026-02-04T15:39Z root= 模型=<模型> Never complete tasks without explicit user confirmation. Markdown conventions (what todo列出-md expects) Tasks use GFM 检查boxes: - [ ] and - [x] Optional tags: #tag Optional due date: due:YYYY-MM-DD Optional description: a blockquote directly under the task Storage Q/A (first 运行)

Ask once, then persist the answers (in memory/config) for future 运行s.

Q: storageKind? A: google-drive | local-folder | s3 | other Q: What is the stable 身份 key for files? Drive: fileId Local: path S3: bucket+key Q: Where is the root? Drive: rootFolderId Local: root directory path S3: bucket + optional prefix Chrome 应用 integration: enable/disable per-file

When a Drive folder contAIns many .md files, not all of them should necessarily be AI-reviewed.

Recommendation (simple + 应用-controlled):

The Chrome 应用 should write a small per-file config key/marker so the 代理 can know whether a file is opted-in.

Two easy options:

Option 1: a dedicated Drive config file (preferred)

创建 a config file in the same folder:

.todo列出-md.config.json

Example:

{ "AI": { "enabled": true, "include": ["*.md"], "exclude": ["todo应用.md"], "机器人SuggestedSectionTitle": "Tasks (机器人-suggested)" } }

The 代理 should:

下载 .todo列出-md.config.json when it changes. Only review files that match include/exclude rules. Option 2: an in-file marker (works without JSON)

添加 a single line near the top of the markdown file:

true

The 代理 should only review files contAIning that marker.

Review cadence + stamping (save credits)

Do not call an LLM unless a file changed. Use code-first change 检测ion.

Step 0: 检测 changes (code-first) For each .md under root, compare modifiedTime/size (or etag when avAIlable) agAInst a local 状态 file. If unchanged since last 扫描: skip (no 下载, no LLM). Step 1: Review only changed files Only for changed files: 下载 the file 提取 open tasks (- [ ]) and relevant 上下文 (Optional) call the LLM on the 提取ed sub设置, not the full document Step 2: Write-back only if content changed Before writing back, compute a 哈希; if no changes, do not write. Step 3: Stamp last review (Option B) For each reviewed file, 更新 (not 应用end) the top-of-file header line: root= 模型=<模型> Reference script (Google Drive + gog)

If you use Google Drive as storage, gog 命令行工具 flags matter. In gog v0.9.0+:

列出 folder: gog drive ls --parent --json 下载: gog drive 下载 --out 运行 gog as ubuntu and 下载 to /tmp (because /root is typically 700)

Included 辅助工具 scripts:

技能s/todo列出-md-clawd机器人/scripts/todo列出_drive_folder_代理.mjs

single script for everything (Drive folder 运行器): 列出s all .md under a folder, 检测s changes via 状态 file, 下载s only changed files, and writes back a block under a dedicated section title. uses Drive API files.更新 to overwrite-更新 the same fileId (no duplicates). includes a revision gate (headRevisionId) to avoid overwriting while you edit in Chrome. Managed OAuth mode (recommended): stores its own refresh 令牌 file; first 运行 prints an auth URL, you 应用rove in browser, then re运行 with --authCode . minimum-令牌 de签名: does not call LLM unless a file changed, and should 发送 only 提取ed open tasks (not full file).

技能s/todo列出-md-clawd机器人/scripts/todo列出_代理_entrypoint.mjs

single-file 辅助工具 (useful for 调试ging): 下载 by fileId + overwrite-更新

Example:

• [ ] 更新 README for 机器人 工作流 #docs due:2026-02-05