Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

🧬 Taiwan.md — Evolve v4.0 — measured AI agent skill | DecimalAI

frank890417/twmd-evolve

NewSkillScoreMaintained

Evolution via canonical EVOLVE-PIPELINE — data-driven content evolution (v1/v2), pipeline self-refactor (Mode 3), goal-driven design evolution (Mode 4: 思考→發散→報告→實作). TRIGGER when: user says "跑 EVOLVE", "數據驅動進化", "evolve-pipeline", "進化 X pipeline 本身", "設計 X（技能/器官）", "深度思考自我進化", "寫實作報告後實作".

agent-toolingby frank890417

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Doc Ops pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/frank890417-twmd-evolve/SKILL.md

decimalai skills pull frank890417-twmd-evolve --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/frank890417-twmd-evolve/SKILL.md https://app.decimal.ai/s/frank890417-twmd-evolve/SKILL.md

Raw file: /s/frank890417-twmd-evolve/SKILL.md ↗ · machine-readable summary at /s/frank890417-twmd-evolve.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 2h ago

Yes — +59% lift+19% tokens0% turns77% of whole cases pass with skill (17/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-02	✗→✓	▲ Improved	3%	0%
case-07	✗→✓	▲ Improved	162%	0%
case-08	✗→✓	▲ Improved	59%	0%
case-09	✗→✓	▲ Improved	11%	0%
case-10	✗→✓	▲ Improved	35%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/3/2026

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

Imported

frank890417 / taiwan-md

.claude/skills/twmd-evolve/SKILL.md

View on GitHub ↗

Imported 8/3/2026Version v1

Skill contents

SKILL.md

🧬 Taiwan.md — Evolve v4.0

🚨 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

Mode 分流（判定寫進 ACK，判準 canonical 在 pipeline）

| 觸發語 | Mode | | ------------------------------------------------------------------ | ------------------------------------------- | | 「跑 EVOLVE」「數據驅動進化」/ twmd-finale 第三棒 / news-lens cron | v1/v2 數據驅動內容進化（default，行為不變） | | 「進化 X pipeline 本身」「pipeline 重組」 | Mode 3 pipeline 自我重組 | | 「設計 X 技能/器官」「深度思考自我進化」「寫實作報告後實作」 | Mode 4 目標驅動設計進化 |

灰區判法：對象是既有 pipeline 自身 = Mode 3；對象是還不存在的能力 = Mode 4。pattern 驅動（DIARY 反覆浮現 ≥3 次）→ 建議改走 /twmd-self-evolve。

Pipeline

嚴格完整讀取並執行 docs/pipelines/EVOLVE-PIPELINE.md v3.6 對應 mode 段落（v1/v2 = Phase 1-7；Mode 3 = 7-stage；Mode 4 = THINK→DIVERGE→REPORT→IMPLEMENT 四相 + 5 hard gate）。

三源交叉 (DNA #4)

GA + SC + CF 三源交叉找放大效應：

GA4 = 誰來了 + 站內行為（page_view / scroll / session_duration）
SC = 誰想來但沒來（query position > 10 + impressions > 100 = 高 demand 低 ranking）
Cloudflare = 誰在邊緣讀我（AI crawler / cached request / 404 rate）

至少 2 源確認的 signal 才升 candidate。

收官

/twmd-finale chain → memory file 必含：BECOME ACK + mode 判定 + Handoff 三態 + Beat 5 反芻。另按 mode 補：v1/v2 = Phase 結果 + 三源 signal 列表 + 選題 candidate；Mode 3 = 重組前後行數對比；Mode 4 = 設計報告 path + 發散方案數 + dogfood 結果。

完整 SOP：docs/pipelines/EVOLVE-PIPELINE.md

Want to see if frank890417/twmd-evolve would activate on your phrasing? Try it before you install — no LLM call, no signup.

Other measured skills in the registry, with their headline benchmark lift.