多元化日常阅读 (Poly Daily Reading)
v1.0.0通过搜索人工智能、LLMs、全栈开发、动漫、恐怖游戏和金属音乐等领域的精选来源，生成Guillaume的每日阅读列表。当要求“生成阅读列表”、“每日阅读”、“找到文章阅读”或运行Poly的每日cron任务时使用。处理去重、Obsidian文件输出、Mission Control导入以及每周/每年的存档。
by 运行时依赖
安装命令
点击复制技能文档
Poly DAIly Reading 列出 (优化d)
生成 and deliver a curated dAIly reading 列出 for 图形界面llaume.
Paths Obsidian vault: ~/库/移动 Documents/iCloud~md~obsidian/Documents/图形界面llaume Maka/ DAIly reading dir: /DAIly Reading/ 输出 file: /DAIly Reading/dAIly-YYYY-MM-DD.md Read-状态 缓存: ~/.OpenClaw/workspace-poly/reading-列出s/read-状态.json Mission Control 命令行工具: ~/.OpenClaw/workspace/mission-control/应用s/命令行工具 工作流
- Collect Past Read URLs (优化d)
Before 搜索ing, gather URLs 图形界面llaume already read from recent 历史:
Glob dAIly-.md files from last 30 days only in DAIly Reading dir 提取 URLs from lines matching [x] (检查ed/read items) Load read-状态.json if it exists for 添加itional 缓存d read URLs Combine into a deduplicated exclusion 设置
辅助工具 script (更新d):
scripts/collect-read-urls.py --days 30
输出s a JSON array of read URLs to stdout.
- 搜索 (优化d)
搜索 across all topics with explicit limits:
Priority order:
Primary AI sources (OpenAI B记录, Anthropic B记录, Kilo.AI B记录, DeepLearning.AI, arxiv) Fetch only latest 5 items per source (not full b记录) Use source-specific feeds/APIs when avAIlable 网页 搜索 for other topics (full-stack, anime, horror games, metal) Limit to 3 结果s per category with count: 3 添加 10-second timeout per 搜索 operation
质量 focus: Still AIm for 2-5 strong picks per category max, but with bounded 搜索 space.
- 过滤器 (Unchanged)
移除 any URL matching the exclusion 设置 from Step 1.
- Write DAIly File (Unchanged)
创建 /DAIly Reading/dAIly-YYYY-MM-DD.md:
# DAIly Reading 列出 — May 8, 2026
AI & 代理ic
- [ ] Title — One-line summary
Full Stack Dev
- [ ] Title — One-line summary
Anime
- [ ] Title — One-line summary
Horror Games
- [ ] Title — One-line summary
Metal Music
- [ ] Title — One-line summary
Categories with no items should still 应用ear (empty section) so 图形界面llaume knows they were 检查ed.
- 更新 Read-状态 缓存 (优化d)
应用end today's new URLs to SINGLE read-状态 file: ~/.OpenClaw/workspace-poly/reading-列出s/read-状态.json
移除 duplicate 更新 to 代理 directory to reduce write operations.
{ "last更新d": "YYYY-MM-DD", "readUrls": ["url1", "url2"] }
- 通知 (Unchanged)
Deliver a brief summary with the link to the Obsidian file.
Mission Control Integration
After saving the dAIly file, ingest into Mission Control:
cd ~/.OpenClaw/workspace/mission-control/应用s/命令行工具 npx tsx src/索引.ts ingest reading-列出 --data ''
JSON 格式化:
{ "代理Id": "poly", "date": "YYYY-MM-DD", "articles": [ {"title": "...", "url": "...", "category": "AI"} ], "delivered": true, "deliveryChannel": "telegram" }
Valid categories: AI, llms, safety, prompting, fullstack, anime, horror, metal
Then 更新 代理 状态:
npx tsx src/索引.ts ingest 状态 --代理-id poly --状态 online --activity-message "Reading 列出 delivered"
Weekly Archiving (Mondays)
Before generating today's 列出, on Mondays:
查询 Mission Control for last week's data: npx tsx src/索引.ts 查询 reading-列出s --dateFrom --dateTo --代理Id poly --格式化 json
Collect all dAIly-.md files from Mon–Sun of last week Group items by source/topic Write to /DAIly Reading/归档s/Month YYYY.md (e.g., "May 2026.md") with header: # Week of May 4-10, 2026
Collected Unread Items
- [ ] Title — Summary
删除 the individual dAIly files after archiving Ingest 归档 task: npx tsx src/索引.ts ingest task --代理Id poly --title "DAIly Reading Weekly 归档" --description "归档d last week's reading 列出s" --状态 completed --category mAIntenance
Yearly Archiving (January 1st)
Before generating today's 列出, on Jan 1:
Collect all dAIly readings from the past year Group by month within a single file Write to /DAIly Reading/YYYY.md (e.g., "2025.md") 删除 monthly 归档 files after creating yearly Ingest 归档 task as above