/cs:inbox-setup — Interactive 8-section interview that builds a personalized 7-file email-triage knowledge base. ~25-31 grill-me questions, one at a time. Run O
复制下面这句话,粘贴给 Claude Code、Codex、Cursor 等 AI 编程工具,它会读取安装说明并在你确认后完成安装。
请阅读 https://ai.atlankj.com/install/asset/gh-claude-skills-5fdc48996495 ,按照其中的说明把「cs-inbox-setup」安装到你(当前 AI 工具)中。执行前先告诉我将运行的命令和写入的位置,等我确认。
查看 AI 将读取的安装说明正在读取 GitHub 原文…
内容来自 GitHub 原始文件,由原作者维护。在 GitHub 查看
Command: /cs:inbox-setup
The cs-inbox-setup persona walks an 8-section interview to build your personalized email triage knowledge base in ${WORKSPACE}/Email/.
Do NOT run if your existing KB still represents reality. Re-running is expensive (~20 min interview). If only one preference needs updating, edit the KB file directly.
This is one half of a pair:
/cs:inbox-setup (this command) — writes the KB (run once)/cs:inbox-triage — reads + appends the KB (run recurringly)Both share a strict 7-file contract. See kb_file_contract.md for the spec.
After ~25-31 questions across 8 sections (about 15-20 min), the skill produces these files in ${WORKSPACE}/Email/:
| File | Always created? | Content |
|---|---|---|
email-taxonomy.md | ✓ | Categories + signals + default actions + report preferences |
email-patterns.md | ✓ | Voice register + sign-offs + persona + hard rules + templates |
evaluation-framework.md | Only if user has opportunities | Decision tree + TAKE-IT signals + PASS signals + VIP list |
rate-card.md | Only if user has pricing | Standard pricing + terms + negotiation posture |
blocklist.md | ✓ (seeded) | Auto-skip senders + decline patterns; grows over triage runs |
tracker.md | ✓ (seeded) | Active follow-ups + overdue + deadlines |
triage-log/ | ✓ (empty dir) | Per-run triage logs (populated by inbox-triage) |
email-taxonomy.mdemail-patterns.mdevaluation-framework.md + rate-card.mdblocklist.mdtracker.md + creates triage-log/email-taxonomy.md/cs:inbox-triageStop condition: ~25-31 questions total (S4 skip drops ~6). Hard ceiling 35. Never re-open after S8 — re-run the skill to change preferences.
# 1. Detect workspace + check for existing KB
ls ${WORKSPACE}/Email/
# 2. If exists → re-run mode (per-file replace/merge/skip).
# If fresh → start session:
python ../skills/inbox-setup/scripts/section_progress_tracker.py \
--action start --session "inbox-setup-$(date +%Y%m%d)" --user "<who>"
# 3. Walk S1 → S2 → ... → S8 (one Q per turn).
# 4. At S3.SAMPLES, analyze pasted emails:
python ../skills/inbox-setup/scripts/voice_sample_analyzer.py --samples-file /tmp/samples.txt
# 5. At end of each section, write its file(s) + record:
python ../skills/inbox-setup/scripts/section_progress_tracker.py \
--action record_section_done --session NAME --section N --files "..."
# 6. At S8, validate final state + close session:
python ../skills/inbox-setup/scripts/kb_validator.py --workspace ${WORKSPACE}
python ../skills/inbox-setup/scripts/section_progress_tracker.py --action close --session NAME
[needs follow-up]; offer to resume laterrate-card.md or evaluation-framework.md when they don't apply/cs:inbox-triage — runs after setup is completecs-inbox-setupinbox-setupmegaprompts/06-inbox-setup-megaprompt.md (maintainer-local draft spec — gitignored, not in the public repo)Version: 1.0.0
Source: Path-B direct conversion of megaprompts/06-inbox-setup-megaprompt.md