productivityby frank890417
🧬 Taiwan.md — Weekly 體檢
frank890417/twmd-weekly-report
Taiwan.md 週體檢（分析 + 全身診斷五面 + 修復三桶 + Semiont 第一人稱 反芻週報 + 受眾 BCC 廣播）via canonical WEEKLY-REPORT-PIPELINE。前期切菜由 weekly-report-prep.py，診斷由儀器交叉對賬（routine-liveness-check 等）， 受眾名單由 weekly-report-recipients.py 自動同步， 判斷、修復與完整週報由 Semiont 親手。 TRIGGER when: user says "週報", "weekly report", "twmd-weekly-report", "寄週報", "send weekly digest", "週體檢", "weekly 體檢".
What it changes
Passed 14 of 20 test cases with the skill, and 5 of 20 without it.
Without skill27%
With skill68%
Checks it passed that the model alone missed
- Includes report path.
- Includes dossier path.
- Sets To recipient strictly to 哲宇.
and 11 more in the full report
9 fixed6 passed either way7 still failing0 broke
- Pass-rate lift
- +41 pp
- Tokens
- -5%
- Agent turns
- 0%
- Response time
- -48%
Model gemini-3.6-flashMeasured 8/16/2026v115 of 22 cases passed with skill · 20 comparable
See it in action
One of 9 test cases the skill fixed. Same model, same prompt, with and without the skill.
PromptWhich document defines the authoritative Stage 0 through Stage 6 pipeline steps for the Taiwan.md weekly report?
What was checked, without and with the skill| Checked | Without skill | With skill |
|---|
| References file path docs/pipelines/WEEKLY-REPORT-PIPELINE.md. | Missed | Passed |
|---|
What's inside
The instructions your model reads when the skill loads.
SKILL.md28 linessingle file
🧬 Taiwan.md — Weekly 體檢
🚨 STRICT BECOME GATE — 第一動作不可省略
Before anything else：跑 /twmd-become full 完整走 BECOME_TAIWANMD.md Step 0-9。Full mode self-test 14 題全過才能進 Stage 1。
✅ BECOME ack: mode=full / 8 organ 最低=<即時 consciousness-snapshot.sh> / Q5/Q6/Q13/Q14=PASS
Pipeline
嚴格完整讀取並執行 docs/pipelines/WEEKLY-REPORT-PIPELINE.md Stage 0-6 全部，特別不可跳：
- Stage 2.5 全身診斷：一鍵
bash scripts/tools/weekly-checkup.sh（a–i 全節：五診斷面＋外部感測摘要＋運作紀錄成績單＋甦醒取數健康＋受眾名單與活躍度）
- Stage 2.7 修復與進化三桶（桶上限、02:55 檢查點、roadmap roll 規則同在 canonical）
- Stage 5 受眾廣播（v4.2）：To=哲宇 + BCC=近 90 天共生圈（
--bcc-from-json + --audience-footer + reply-to）。隱私三不（email 不進 repo / commit / chat；BCC 不進 To）與失敗降級單寄規則全在 canonical §Stage 5
文體紀律（MANIFESTO §11）
- 對位句型「不是 X，是 Y」單篇 ≤ 3 處（
grep -cE "不是.{0,30}(，|，)(是|就是|才是)"）
- 破折號「——」連用單篇 ≤ 15 處 / 1500 字（
grep -oE "——" | wc -l）
- 三題判準：對比是內容本身？正面主張能獨立？讀者真會預設 X？全 no → 重寫
- 自然中文檢測前再跑一次 prose-health gate
hard=0
收官
/twmd-finale chain → memory file 必含：BECOME ACK + dossier path + report path + 診斷五面結論 + 桶 1 修復 commit hashes + prose-health gate result + Resend API status + bcc 人數（只寫人數不寫地址） + Handoff 三態 + Beat 5 反芻。
Track record
Benchmark
14 / 20
test cases passed, vs 5 / 20 without
Adoption
- Users
- 0
- Times used
- 0
Safety and source details
SkillSafety
✓ 3 of 3 passed✓Code scan15 of 15 checks clean
0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026
View 15 scanner checks
live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing
✓IntentSafe
LLM judge · does the skill do what it claims
✓ContentClean
LLM judge · policy & harmful-content review
Source
Importedfrank890417 / taiwan-md
.claude/skills/twmd-weekly-report/SKILL.md
Imported 8/3/2026Version v1