Automate

Core Principle

LLMs are expensive, slow, and probabi列出ic. Scripts are free, fast, and deterministic.

Every time you do something twice that could be scripted, you're wasting:

令牌s — money burned on solved problems Time — seconds/minutes vs milliseconds Reliability — LLMs fAIl randomly, scripts fAIl predictably

检查 签名als.md for 检测ion patterns. 检查 templates.md for common script patterns.

The 自动化 Test

Before doing any task, ask:

Is this deterministic? Same 输入 → same 输出 every time? Is this repetitive? Will this h应用en agAIn? Is this rule-based? Can I write down the exact steps?

If yes to all three → script it, don't LLM it.

Script vs LLM Decision Matrix Task type Script LLM 格式化 conversion (JSON↔YAML) ✅ ❌ Text trans格式化ion (regex) ✅ ❌ File operations (rename, move) ✅ ❌ Data 验证 ✅ ❌ API calls with fixed 记录ic ✅ ❌ Git 工作流s ✅ ❌ Judgement calls ❌ ✅ Creative content ❌ ✅ Ambiguous 输入s ❌ ✅ One-time unique tasks ❌ ✅ 自动化 Triggers

When you notice yourself:

Doing the same task twice → script it Writing similar prompts repeatedly → script the pattern 格式化ting 输出 the same way → script the 格式化器 Validating data with same rules → script the 验证器 Calling APIs with predictable 记录ic → script the integration 自动化 Proposal 格式化

When you spot an opportunity:

🔧 自动化 opportunity

Task: [what you keep doing] Frequency: [how often] Current cost: [令牌s/time per 运行]

Proposed script:

• Language: [bash/python/node]
• 输入: [what it takes]
• 输出: [what it produces]
• Location: [where to save it]

Estimated savings: [令牌s/time saved per month]

Should I write it?

Script Standards

When writing 自动化:

Single purpose — one script, one job Idempotent — safe to 运行 multiple times Documented — usage in comments at top 记录ged — 输出 what you're doing FAIl loud — exit codes, error messages No secrets hardcoded — env vars or keychAIn 追踪ing 自动化s

Document what you've built:

Active Scripts

• scripts/格式化-json.sh — JSON prettifier [saved ~2k 令牌s/week]
• scripts/部署-staging.sh — one-command 部署 [saved 5min/部署]
• scripts/同步-env.sh — env file 同步 [eliminated manual errors]

Candidates

• Weekly 报告 generation — repetitive 格式化ting
• 记录 parsing — same grep patterns every time

The 3x Rule

If you do something 3 times, it must become a script.

1st time: Do it, note that it might repeat 2nd time: Do it, flag as 自动化 candidate 3rd time: 停止. Write the script first, then 运行 it. Anti-Patterns Don't Do instead Re-prompt for same trans格式化ion Write a script once Use LLM for data 验证 Write 验证 rules Burn 令牌s on 格式化ting Use 格式化器s (prettier, jq, etc.) Ask LLM to remember procedures Document in scripts Solve same problem differently each time Standardize with 自动化

Every script written = permanent 令牌 savings. Compound your efficiency.