[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6MG72WQRHSPWVT0RGC4C7B6":3,"skill-same:01M6MG72WQRHSPWVT0RGC4C7B6":76},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":25,"publishedAt":46,"linkOnly":47,"markdownContent":48,"markdownBody":49,"frontMatter":50,"refName":56,"commitSha":57,"skillDirectory":58,"forksCount":13,"repositoryHtmlUrl":59,"repositoryHomepageUrl":6,"ownerHtmlUrl":60,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":61,"install":62,"resources":75},"01M6MG72WQRHSPWVT0RGC4C7B6","inbox-triage",null,"Runs a full inbox triage using the knowledge base created by the 'inbox-setup' skill. Light-intake by design (most invocations skip questions and run with KB-default preferences); asks at most 2 grill-me override questions when invocation is outside normal cadence or includes category-skip intent. Searches recent emails, classifies them via the user's taxonomy, researches new senders, generates recommendations, drafts replies (NEVER sends), delivers a report in the user's preferred format, and updates the knowledge base with learnings. Designed to run on a recurring schedule (1-3x daily) or on demand. Use when the user wants their inbox processed, in any variation (e.g., 'triage my inbox', 'inbox triage', 'check my email', 'run email triage', 'process my inbox', 'what's new in my email', 'handle my email', 'email triage'). Requires the inbox-setup skill to have been run first.","JiayiV\u002Fquestion_mixing","JiayiV","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F54466889?v=4","01M6JFJYYMRZ3QRGP9339JNQ2K","01M6JFJZ35KB0QVT50NAHPJA7G",0,"MIT","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fpublic-02-claude-skills-main-gemini-skills-inbox-triage-skill-e510c8d8\u002FSKILL.md",[17,22],{"categoryNo":18,"categoryName":19,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000005","邮件与沟通","SCENARIO","使用场景",{"categoryNo":23,"categoryName":24,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000021","资料检索与研究",[26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45],"triage","inbox triage","inbox","updates","design","cadence","recommendations","knowledge base","me","learnings","intent","grill","default","conversion","grill-me","never","kb","schedule","knowledge","run","2026-11-10 10:12:30",false,"---\nname: inbox-triage\ndescription: \"Runs a full inbox triage using the knowledge base created by the 'inbox-setup' skill. Light-intake by design (most invocations skip questions and run with KB-default preferences); asks at most 2 grill-me override questions when invocation is outside normal cadence or includes category-skip intent. Searches recent emails, classifies them via the user's taxonomy, researches new senders, generates recommendations, drafts replies (NEVER sends), delivers a report in the user's preferred format, and updates the knowledge base with learnings. Designed to run on a recurring schedule (1-3x daily) or on demand. Use when the user wants their inbox processed, in any variation (e.g., 'triage my inbox', 'inbox triage', 'check my email', 'run email triage', 'process my inbox', 'what's new in my email', 'handle my email', 'email triage'). Requires the inbox-setup skill to have been run first.\"\nlicense: MIT\nmetadata:\n  source_spec: \"megaprompts\u002F07-inbox-triage-megaprompt.md\"\n  build_pattern: \"Path B (direct conversion)\"\n  paired_with: \"inbox-setup (consumes the 7-file KB it produces)\"\n  version: 1.0.0\n---\n\n# Inbox-Triage — Recurring Email Triage\n\n> **Paired with `inbox-setup`.** This skill consumes the 7-file knowledge base that `inbox-setup` writes at `${WORKSPACE}\u002FEmail\u002F`. The file contracts MUST match exactly. See [`references\u002Fkb_file_contract.md`](references\u002Fkb_file_contract.md) — this is the mirror of the setup-side contract, viewed from the read side.\n\nRun on a recurring schedule (1–3x daily) or on demand. Classify recent emails, research new senders, generate decision recommendations, draft replies (**NEVER SEND**), deliver a clean report, and update the knowledge base with what was learned this run.\n\n## Invocation Triggers\n\n- \"triage my inbox\"\n- \"inbox triage\"\n- \"check my email\"\n- \"run email triage\"\n- \"process my inbox\"\n- \"what's new in my email\"\n- \"handle my email\"\n- \"email triage\"\n\n## Prerequisites\n\nRequired reads at start (fail-fast if missing):\n\n**Core (required):**\n- `${WORKSPACE}\u002FEmail\u002Femail-taxonomy.md` — classification + report preferences\n- `${WORKSPACE}\u002FEmail\u002Femail-patterns.md` — voice, persona, templates, hard rules\n\n**Optional core (read if exists):**\n- `${WORKSPACE}\u002FEmail\u002Fevaluation-framework.md`\n- `${WORKSPACE}\u002FEmail\u002Frate-card.md`\n\n**Evolving (read AND update every run):**\n- `${WORKSPACE}\u002FEmail\u002Fblocklist.md`\n- `${WORKSPACE}\u002FEmail\u002Ftracker.md`\n\n**Output:**\n- `${WORKSPACE}\u002FEmail\u002Ftriage-log\u002F\u003CYYYY-MM-DD>-\u003Crun-label>.md` — per-run log\n\nIf any core required file is missing → **halt**, direct user to run `inbox-setup` first. Use `scripts\u002Fkb_reader.py` to perform the read + validation.\n\n## DRAFTS ONLY — Never Send\n\n> **This skill creates drafts. It NEVER sends.**\n\nThis is the safety property that makes the skill safe to run automatically. Stated multiple times in this skill body. Non-negotiable.\n\nThe `scripts\u002Fdraft_safety_validator.py` enforces it post-run. Any send-shaped tool call in the action log fails validation. See [`references\u002Fdrafts_only_safety.md`](references\u002Fdrafts_only_safety.md) for the full discipline canon.\n\n## Step 0: Grill-Me Intake (Light — 0–2 Optional Override Questions)\n\nInbox-triage is **light-intake by design** — it runs on a recurring cadence with preferences pre-baked into the knowledge base from `inbox-setup`. The grill-me discipline here is asking ONLY the override questions that matter THIS run.\n\n### Q1 (optional, asked only when on-demand run is outside normal cadence)\n\n> **Override the default 9-hour search window? Pick: yes (specify hours) \u002F no (use default).**\n>\n> *Why I'm asking:* If you're running on-demand outside your normal 2x\u002Fday cadence, you may want a wider window (24h after a long break) or narrower (2h for a quick check).\n\nSkip if cadence is normal.\n\n### Q2 (optional, asked only when user invokes with category-skip intent)\n\n> **Skip any categories this run? E.g., \"skip newsletters\", \"skip financial\".**\n>\n> *Why I'm asking:* Sometimes you just want to scan opportunities or just want to clear active threads. Category skip narrows the run scope.\n\nSkip if user gave no category-skip signal.\n\n**Stop condition:** Max 2 questions. Default invocations skip both questions and run with KB-default preferences. The skill is optimized for fast recurring execution; intake is the exception, not the norm.\n\n## Step 1: Determine Search Window\n\nCompute via current date math. Default lookback: **9 hours** (works for 2x\u002Fday cadence with slight overlap so emails between runs aren't missed).\n\nUse `scripts\u002Fsearch_window_calculator.py --cadence \u003CCADENCE> --now \u003CISO>`:\n\n```\nnow = current_datetime\nwindow_start = now - 9_hours   (default for 2x-daily)\nrun_label = \"Morning\" if now.hour \u003C 12 else \"Afternoon\" if now.hour \u003C 17 else \"Evening\"\n```\n\nCadence-to-default-window mapping (override via Q1):\n\n| Cadence (from email-taxonomy.md S1.Q5) | Default window |\n|---|---|\n| once daily | 26h |\n| 2x daily | 9h |\n| 3x daily | 6h |\n| on-demand only | 24h (asks Q1) |\n\n## Step 2: Email Search\n\nTwo queries (provider-agnostic adapter pattern):\n\n- **Primary:** Inbox + sent after `window_start`\n- **Secondary:** Starred unread (catch flagged items missed in primary)\n\nCollect for each email: sender, subject, date, snippet, thread ID, labels.\n\nProvider adapter mapping:\n\n| Provider | Tool |\n|---|---|\n| Gmail | Gmail MCP |\n| Outlook \u002F Microsoft 365 | Outlook MCP |\n| IMAP (Fastmail, ProtonMail, etc.) | IMAP MCP if available; halt otherwise |\n| (no email tool available) | Halt with clear message: \"No email tool registered for this session.\" |\n\n## Step 3: Classification\n\nApply the taxonomy from `email-taxonomy.md`. For **lowest-priority** category (newsletters \u002F automation \u002F spam): skip thread reads entirely — context cost not worth it. For everything else: read full thread.\n\n## Step 4: Sender Research\n\nFor senders not in tracker \u002F blocklist \u002F prior logs:\n\n1. Check `blocklist.md` → if matched, auto-skip\n2. Check `tracker.md` → if known thread, note existing context\n3. For opportunity senders (per evaluation framework): web search for company legitimacy, social presence, intermediary status\n\n**Skip research entirely** for: known senders (in tracker), internal email, automated notifications, obvious low-priority.\n\n## Step 5: Recommendations\n\nFor decision-required emails, apply the framework from `evaluation-framework.md`. Categorize:\n\n| Category | When | Output |\n|---|---|---|\n| **TAKE IT** | Meets criteria | Recommend engaging; draft reply (Step 6) |\n| **WORTH CONSIDERING** | Has potential, needs user judgment | Surface key context; draft for user to edit |\n| **PASS** | Doesn't meet criteria | Brief \"why\" (1–3 sentences); draft polite decline |\n| **FLAG FOR REVIEW** | Unusual; needs direct user decision | Surface fully; NO draft (user decides response shape) |\n\nEach: brief \"why\", relevant context, pricing\u002Ftimeline comparison if applicable.\n\n**Skip Step 5 entirely if no `evaluation-framework.md` exists.**\n\nSee [`references\u002Ftriage_decision_framework.md`](references\u002Ftriage_decision_framework.md) for the framework canon.\n\n## Step 6: Drafts\n\nFor every reasonable reply candidate, create a draft using `email-patterns.md` voice rules.\n\n**Draft for:** opportunity responses (TAKE IT \u002F WORTH \u002F PASS), active conversations needing reply, action items, important personal emails.\n\n**Do NOT draft for:**\n- Clearly no-response emails (newsletters, automation, FYI)\n- Threads where user already replied\n- Blocked senders (unless new info changes the calculus)\n\n**Mechanics:**\n\n- Draft only in the existing thread when possible (preserves context)\n- Set `to`, `subject` (`Re: [original]`)\n- **NEVER call any send operation. Only create drafts.**\n\nThe draft body MUST honor:\n- Voice register from `email-patterns.md`\n- Forbidden tokens (S3.Q2 pet peeves)\n- Sign-off patterns\n- Persona context\n- Hard rules (S3.Q6 — non-negotiable)\n- Reply length per `email-patterns.md`\n\nIf `evaluation-framework.md` exists, draft tone matches recommendation:\n- TAKE IT → engaged + concrete next step\n- WORTH → curious + 1-2 clarifying questions\n- PASS → polite decline + brief reason (no hedging promises)\n- FLAG → NO draft\n\n## Step 7: Report Delivery\n\nHonor user's preference from `email-taxonomy.md` \"Report Preferences\" section. Default: email draft to self with HTML.\n\n**Subject:** `Inbox Triage — [Day], [Month Date] ([Run Label])`\n\n**Sections (in order):**\n\n1. **Overview** — 2–3 sentences. What happened? Anything urgent?\n2. **Stats** — Counts: processed, drafts created, action needed, skipped.\n3. **Action Needed** — Overdue items, decisions, drafts to review, deadlines.\n4. **Quick Reference** — One line per email, alphabetical by sender. `**Sender** — one-sentence summary + recommendation`.\n5. **Detailed Cards** — Opportunities, active threads, flags. Each: sender \u002F subject \u002F category, recommendation + reasoning, key context. **NO draft text previews** (drafts are already in email client for user to read there).\n6. **Footer** — Generation timestamp + KB update summary.\n\n**Formatting (if HTML):**\n\n- **Inline CSS only** (Gmail strips `\u003Cstyle>`)\n- Color-coded by recommendation:\n  - green → TAKE IT\n  - amber → WORTH CONSIDERING\n  - red → PASS\n  - purple → FLAG FOR REVIEW\n  - blue → active conversation\n\n## Step 8: Knowledge Base Update\n\n**`blocklist.md`** (append new):\n\n- New declined senders + reason + date\n- New decline patterns from observed behavior (e.g., \"all emails containing 'looking for backend engineers' from gmail addresses → cold recruiter pattern\")\n- Remove entries if user has overridden them (user replied to a \"blocked\" sender → unblock)\n\n**`tracker.md`** (append + update):\n\n- New follow-ups for emails needing future action\n- Update existing follow-ups (deadline changed, status changed)\n- Mark resolved items complete\n- Flag overdue items\n- Remove resolved items older than 30 days\n- Add entry to update log\n\n**Learning patterns to observe over runs:**\n\n- Drafts sent as-is vs. edited vs. deleted → tone calibration signal\n- PASS recommendations user overrides → framework adjustment signal\n- Engaged vs. ignored emails → taxonomy refinement signal\n- New decline patterns → blocklist additions\n\nAfter 5+ runs, suggest KB improvements to user (e.g., \"You always decline emails from X — add as auto-skip?\").\n\n## Step 9: Internal Log\n\nSave to `${WORKSPACE}\u002FEmail\u002Ftriage-log\u002F[YYYY-MM-DD]-[run-label].md`:\n\n- Emails processed with classifications\n- Recommendations made\n- Drafts created (with IDs \u002F thread refs)\n- KB updates made\n- Follow-ups added \u002F resolved\n- Notable observations (patterns surfaced, edge cases handled)\n\nThe log is the audit trail for `scripts\u002Fdraft_safety_validator.py` to scan for send operations post-run.\n\n## Step 10: Empty Inbox Handling\n\nEven with zero new emails:\n\n1. Check `tracker.md` for items due today or overdue\n2. Generate minimal report: \"No new actionable emails since last run\"\n3. Flag any overdue items\n4. Escalate per tracker rules\n\nSkip Steps 3–6 entirely on empty inbox.\n\n## Critical Rules (Stated Multiple Times)\n\n1. **DRAFTS ONLY — NEVER SEND.** Non-negotiable. Stated again here.\n2. **Privacy.** No passwords \u002F credentials in KB. Reference threads by ID for sensitive content.\n3. **Accuracy over speed.** When unsure, flag for review. A wrong auto-draft is worse than no draft.\n4. **Respect the KB.** Documented preferences are source of truth. Don't override with judgment.\n5. **Transparency.** Note every KB change in the triage log.\n6. **First runs need oversight.** Document this expectation for the user.\n\n## Error Handling\n\n| Situation | Behavior |\n|---|---|\n| KB files missing | Halt; direct user to run `inbox-setup` |\n| Email tool unavailable | Halt with clear message about required tool |\n| Web search unavailable for sender research | Skip research step; note senders not researched |\n| Draft creation fails | Skip that draft; note in log; report continues |\n| Report delivery fails | Save report to file as fallback; notify user |\n| User has 100+ new emails | Stay within reasonable limits; flag volume; offer to focus on priority categories only |\n| Sender appears in both blocklist and tracker | Tracker wins (active conversation); note inconsistency in log |\n\n## Portability\n\n- **Claude Code CLI:** Native — uses Gmail \u002F Outlook MCP, file tools for KB, web search for research.\n- **Claude.ai web:** Works when email MCP connector is connected (Gmail MCP available). Skill must check tool availability before assuming. If no email tool: halt with clear message.\n\n## Tooling\n\n| Script | Role |\n|---|---|\n| `scripts\u002Fkb_reader.py` | Reads + validates the 7-file KB. Returns parsed structure. Halts with explicit error if required files missing. |\n| `scripts\u002Fsearch_window_calculator.py` | Computes `window_start` from cadence + current time. Returns `run_label`. Honors Q1 override. |\n| `scripts\u002Fdraft_safety_validator.py` | Post-run scan of the action log for any send-shaped tool call. FAILs if detected. The deterministic enforcement of the NEVER-SEND rule. |\n\n## References\n\n- [`references\u002Fkb_file_contract.md`](references\u002Fkb_file_contract.md) — canonical 7-file contract (read perspective; mirrors `..\u002Finbox-setup\u002Freferences\u002Fkb_file_contract.md`)\n- [`references\u002Ftriage_decision_framework.md`](references\u002Ftriage_decision_framework.md) — TAKE IT \u002F WORTH \u002F PASS \u002F FLAG taxonomy\n- [`references\u002Fdrafts_only_safety.md`](references\u002Fdrafts_only_safety.md) — the NEVER-SEND discipline canon\n\n## Anti-Patterns To Reject\n\n- **Sending emails** (drafts only — non-negotiable)\n- Operating without knowledge base files\n- Storing passwords \u002F credentials in KB\n- Skipping the learning loop (KB updates) at end of run\n- Overriding user's documented preferences with own judgment\n- Reading lowest-priority threads (waste of context)\n- Including draft text previews in report (drafts are already in email client)\n- Provider lock-in without adapter pattern\n- Silently failing on missing tools\n\n---\n\n**Version:** 1.0.0\n**Source spec:** `megaprompts\u002F07-inbox-triage-megaprompt.md` (maintainer-local draft spec — gitignored, not present in the public repository)\n**Build pattern:** Path B (direct conversion). Paired with `inbox-setup`.\n","# Inbox-Triage — Recurring Email Triage\n\n> **Paired with `inbox-setup`.** This skill consumes the 7-file knowledge base that `inbox-setup` writes at `${WORKSPACE}\u002FEmail\u002F`. The file contracts MUST match exactly. See [`references\u002Fkb_file_contract.md`](references\u002Fkb_file_contract.md) — this is the mirror of the setup-side contract, viewed from the read side.\n\nRun on a recurring schedule (1–3x daily) or on demand. Classify recent emails, research new senders, generate decision recommendations, draft replies (**NEVER SEND**), deliver a clean report, and update the knowledge base with what was learned this run.\n\n## Invocation Triggers\n\n- \"triage my inbox\"\n- \"inbox triage\"\n- \"check my email\"\n- \"run email triage\"\n- \"process my inbox\"\n- \"what's new in my email\"\n- \"handle my email\"\n- \"email triage\"\n\n## Prerequisites\n\nRequired reads at start (fail-fast if missing):\n\n**Core (required):**\n- `${WORKSPACE}\u002FEmail\u002Femail-taxonomy.md` — classification + report preferences\n- `${WORKSPACE}\u002FEmail\u002Femail-patterns.md` — voice, persona, templates, hard rules\n\n**Optional core (read if exists):**\n- `${WORKSPACE}\u002FEmail\u002Fevaluation-framework.md`\n- `${WORKSPACE}\u002FEmail\u002Frate-card.md`\n\n**Evolving (read AND update every run):**\n- `${WORKSPACE}\u002FEmail\u002Fblocklist.md`\n- `${WORKSPACE}\u002FEmail\u002Ftracker.md`\n\n**Output:**\n- `${WORKSPACE}\u002FEmail\u002Ftriage-log\u002F\u003CYYYY-MM-DD>-\u003Crun-label>.md` — per-run log\n\nIf any core required file is missing → **halt**, direct user to run `inbox-setup` first. Use `scripts\u002Fkb_reader.py` to perform the read + validation.\n\n## DRAFTS ONLY — Never Send\n\n> **This skill creates drafts. It NEVER sends.**\n\nThis is the safety property that makes the skill safe to run automatically. Stated multiple times in this skill body. Non-negotiable.\n\nThe `scripts\u002Fdraft_safety_validator.py` enforces it post-run. Any send-shaped tool call in the action log fails validation. See [`references\u002Fdrafts_only_safety.md`](references\u002Fdrafts_only_safety.md) for the full discipline canon.\n\n## Step 0: Grill-Me Intake (Light — 0–2 Optional Override Questions)\n\nInbox-triage is **light-intake by design** — it runs on a recurring cadence with preferences pre-baked into the knowledge base from `inbox-setup`. The grill-me discipline here is asking ONLY the override questions that matter THIS run.\n\n### Q1 (optional, asked only when on-demand run is outside normal cadence)\n\n> **Override the default 9-hour search window? Pick: yes (specify hours) \u002F no (use default).**\n>\n> *Why I'm asking:* If you're running on-demand outside your normal 2x\u002Fday cadence, you may want a wider window (24h after a long break) or narrower (2h for a quick check).\n\nSkip if cadence is normal.\n\n### Q2 (optional, asked only when user invokes with category-skip intent)\n\n> **Skip any categories this run? E.g., \"skip newsletters\", \"skip financial\".**\n>\n> *Why I'm asking:* Sometimes you just want to scan opportunities or just want to clear active threads. Category skip narrows the run scope.\n\nSkip if user gave no category-skip signal.\n\n**Stop condition:** Max 2 questions. Default invocations skip both questions and run with KB-default preferences. The skill is optimized for fast recurring execution; intake is the exception, not the norm.\n\n## Step 1: Determine Search Window\n\nCompute via current date math. Default lookback: **9 hours** (works for 2x\u002Fday cadence with slight overlap so emails between runs aren't missed).\n\nUse `scripts\u002Fsearch_window_calculator.py --cadence \u003CCADENCE> --now \u003CISO>`:\n\n```\nnow = current_datetime\nwindow_start = now - 9_hours   (default for 2x-daily)\nrun_label = \"Morning\" if now.hour \u003C 12 else \"Afternoon\" if now.hour \u003C 17 else \"Evening\"\n```\n\nCadence-to-default-window mapping (override via Q1):\n\n| Cadence (from email-taxonomy.md S1.Q5) | Default window |\n|---|---|\n| once daily | 26h |\n| 2x daily | 9h |\n| 3x daily | 6h |\n| on-demand only | 24h (asks Q1) |\n\n## Step 2: Email Search\n\nTwo queries (provider-agnostic adapter pattern):\n\n- **Primary:** Inbox + sent after `window_start`\n- **Secondary:** Starred unread (catch flagged items missed in primary)\n\nCollect for each email: sender, subject, date, snippet, thread ID, labels.\n\nProvider adapter mapping:\n\n| Provider | Tool |\n|---|---|\n| Gmail | Gmail MCP |\n| Outlook \u002F Microsoft 365 | Outlook MCP |\n| IMAP (Fastmail, ProtonMail, etc.) | IMAP MCP if available; halt otherwise |\n| (no email tool available) | Halt with clear message: \"No email tool registered for this session.\" |\n\n## Step 3: Classification\n\nApply the taxonomy from `email-taxonomy.md`. For **lowest-priority** category (newsletters \u002F automation \u002F spam): skip thread reads entirely — context cost not worth it. For everything else: read full thread.\n\n## Step 4: Sender Research\n\nFor senders not in tracker \u002F blocklist \u002F prior logs:\n\n1. Check `blocklist.md` → if matched, auto-skip\n2. Check `tracker.md` → if known thread, note existing context\n3. For opportunity senders (per evaluation framework): web search for company legitimacy, social presence, intermediary status\n\n**Skip research entirely** for: known senders (in tracker), internal email, automated notifications, obvious low-priority.\n\n## Step 5: Recommendations\n\nFor decision-required emails, apply the framework from `evaluation-framework.md`. Categorize:\n\n| Category | When | Output |\n|---|---|---|\n| **TAKE IT** | Meets criteria | Recommend engaging; draft reply (Step 6) |\n| **WORTH CONSIDERING** | Has potential, needs user judgment | Surface key context; draft for user to edit |\n| **PASS** | Doesn't meet criteria | Brief \"why\" (1–3 sentences); draft polite decline |\n| **FLAG FOR REVIEW** | Unusual; needs direct user decision | Surface fully; NO draft (user decides response shape) |\n\nEach: brief \"why\", relevant context, pricing\u002Ftimeline comparison if applicable.\n\n**Skip Step 5 entirely if no `evaluation-framework.md` exists.**\n\nSee [`references\u002Ftriage_decision_framework.md`](references\u002Ftriage_decision_framework.md) for the framework canon.\n\n## Step 6: Drafts\n\nFor every reasonable reply candidate, create a draft using `email-patterns.md` voice rules.\n\n**Draft for:** opportunity responses (TAKE IT \u002F WORTH \u002F PASS), active conversations needing reply, action items, important personal emails.\n\n**Do NOT draft for:**\n- Clearly no-response emails (newsletters, automation, FYI)\n- Threads where user already replied\n- Blocked senders (unless new info changes the calculus)\n\n**Mechanics:**\n\n- Draft only in the existing thread when possible (preserves context)\n- Set `to`, `subject` (`Re: [original]`)\n- **NEVER call any send operation. Only create drafts.**\n\nThe draft body MUST honor:\n- Voice register from `email-patterns.md`\n- Forbidden tokens (S3.Q2 pet peeves)\n- Sign-off patterns\n- Persona context\n- Hard rules (S3.Q6 — non-negotiable)\n- Reply length per `email-patterns.md`\n\nIf `evaluation-framework.md` exists, draft tone matches recommendation:\n- TAKE IT → engaged + concrete next step\n- WORTH → curious + 1-2 clarifying questions\n- PASS → polite decline + brief reason (no hedging promises)\n- FLAG → NO draft\n\n## Step 7: Report Delivery\n\nHonor user's preference from `email-taxonomy.md` \"Report Preferences\" section. Default: email draft to self with HTML.\n\n**Subject:** `Inbox Triage — [Day], [Month Date] ([Run Label])`\n\n**Sections (in order):**\n\n1. **Overview** — 2–3 sentences. What happened? Anything urgent?\n2. **Stats** — Counts: processed, drafts created, action needed, skipped.\n3. **Action Needed** — Overdue items, decisions, drafts to review, deadlines.\n4. **Quick Reference** — One line per email, alphabetical by sender. `**Sender** — one-sentence summary + recommendation`.\n5. **Detailed Cards** — Opportunities, active threads, flags. Each: sender \u002F subject \u002F category, recommendation + reasoning, key context. **NO draft text previews** (drafts are already in email client for user to read there).\n6. **Footer** — Generation timestamp + KB update summary.\n\n**Formatting (if HTML):**\n\n- **Inline CSS only** (Gmail strips `\u003Cstyle>`)\n- Color-coded by recommendation:\n  - green → TAKE IT\n  - amber → WORTH CONSIDERING\n  - red → PASS\n  - purple → FLAG FOR REVIEW\n  - blue → active conversation\n\n## Step 8: Knowledge Base Update\n\n**`blocklist.md`** (append new):\n\n- New declined senders + reason + date\n- New decline patterns from observed behavior (e.g., \"all emails containing 'looking for backend engineers' from gmail addresses → cold recruiter pattern\")\n- Remove entries if user has overridden them (user replied to a \"blocked\" sender → unblock)\n\n**`tracker.md`** (append + update):\n\n- New follow-ups for emails needing future action\n- Update existing follow-ups (deadline changed, status changed)\n- Mark resolved items complete\n- Flag overdue items\n- Remove resolved items older than 30 days\n- Add entry to update log\n\n**Learning patterns to observe over runs:**\n\n- Drafts sent as-is vs. edited vs. deleted → tone calibration signal\n- PASS recommendations user overrides → framework adjustment signal\n- Engaged vs. ignored emails → taxonomy refinement signal\n- New decline patterns → blocklist additions\n\nAfter 5+ runs, suggest KB improvements to user (e.g., \"You always decline emails from X — add as auto-skip?\").\n\n## Step 9: Internal Log\n\nSave to `${WORKSPACE}\u002FEmail\u002Ftriage-log\u002F[YYYY-MM-DD]-[run-label].md`:\n\n- Emails processed with classifications\n- Recommendations made\n- Drafts created (with IDs \u002F thread refs)\n- KB updates made\n- Follow-ups added \u002F resolved\n- Notable observations (patterns surfaced, edge cases handled)\n\nThe log is the audit trail for `scripts\u002Fdraft_safety_validator.py` to scan for send operations post-run.\n\n## Step 10: Empty Inbox Handling\n\nEven with zero new emails:\n\n1. Check `tracker.md` for items due today or overdue\n2. Generate minimal report: \"No new actionable emails since last run\"\n3. Flag any overdue items\n4. Escalate per tracker rules\n\nSkip Steps 3–6 entirely on empty inbox.\n\n## Critical Rules (Stated Multiple Times)\n\n1. **DRAFTS ONLY — NEVER SEND.** Non-negotiable. Stated again here.\n2. **Privacy.** No passwords \u002F credentials in KB. Reference threads by ID for sensitive content.\n3. **Accuracy over speed.** When unsure, flag for review. A wrong auto-draft is worse than no draft.\n4. **Respect the KB.** Documented preferences are source of truth. Don't override with judgment.\n5. **Transparency.** Note every KB change in the triage log.\n6. **First runs need oversight.** Document this expectation for the user.\n\n## Error Handling\n\n| Situation | Behavior |\n|---|---|\n| KB files missing | Halt; direct user to run `inbox-setup` |\n| Email tool unavailable | Halt with clear message about required tool |\n| Web search unavailable for sender research | Skip research step; note senders not researched |\n| Draft creation fails | Skip that draft; note in log; report continues |\n| Report delivery fails | Save report to file as fallback; notify user |\n| User has 100+ new emails | Stay within reasonable limits; flag volume; offer to focus on priority categories only |\n| Sender appears in both blocklist and tracker | Tracker wins (active conversation); note inconsistency in log |\n\n## Portability\n\n- **Claude Code CLI:** Native — uses Gmail \u002F Outlook MCP, file tools for KB, web search for research.\n- **Claude.ai web:** Works when email MCP connector is connected (Gmail MCP available). Skill must check tool availability before assuming. If no email tool: halt with clear message.\n\n## Tooling\n\n| Script | Role |\n|---|---|\n| `scripts\u002Fkb_reader.py` | Reads + validates the 7-file KB. Returns parsed structure. Halts with explicit error if required files missing. |\n| `scripts\u002Fsearch_window_calculator.py` | Computes `window_start` from cadence + current time. Returns `run_label`. Honors Q1 override. |\n| `scripts\u002Fdraft_safety_validator.py` | Post-run scan of the action log for any send-shaped tool call. FAILs if detected. The deterministic enforcement of the NEVER-SEND rule. |\n\n## References\n\n- [`references\u002Fkb_file_contract.md`](references\u002Fkb_file_contract.md) — canonical 7-file contract (read perspective; mirrors `..\u002Finbox-setup\u002Freferences\u002Fkb_file_contract.md`)\n- [`references\u002Ftriage_decision_framework.md`](references\u002Ftriage_decision_framework.md) — TAKE IT \u002F WORTH \u002F PASS \u002F FLAG taxonomy\n- [`references\u002Fdrafts_only_safety.md`](references\u002Fdrafts_only_safety.md) — the NEVER-SEND discipline canon\n\n## Anti-Patterns To Reject\n\n- **Sending emails** (drafts only — non-negotiable)\n- Operating without knowledge base files\n- Storing passwords \u002F credentials in KB\n- Skipping the learning loop (KB updates) at end of run\n- Overriding user's documented preferences with own judgment\n- Reading lowest-priority threads (waste of context)\n- Including draft text previews in report (drafts are already in email client)\n- Provider lock-in without adapter pattern\n- Silently failing on missing tools\n\n---\n\n**Version:** 1.0.0\n**Source spec:** `megaprompts\u002F07-inbox-triage-megaprompt.md` (maintainer-local draft spec — gitignored, not present in the public repository)\n**Build pattern:** Path B (direct conversion). Paired with `inbox-setup`.\n",{"name":5,"license":14,"metadata":51,"description":7},{"version":52,"paired_with":53,"source_spec":54,"build_pattern":55},"1.0.0","inbox-setup (consumes the 7-file KB it produces)","megaprompts\u002F07-inbox-triage-megaprompt.md","Path B (direct conversion)","main","068c2486ccc0f8d200a3db4d21ec137064eda3b4","skills\u002Fpublic-02-claude-skills-main-gemini-skills-inbox-triage-skill-e510c8d8","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing","https:\u002F\u002Fgithub.com\u002FJiayiV","2026-08-25 16:30:52",{"source":8,"skillName":5,"commands":63,"prompt":73,"zipAvailable":74},[64,67,70],{"tool":65,"command":66},"npx","npx skills add JiayiV\u002Fquestion_mixing --skill inbox-triage",{"tool":68,"command":69},"bunx","bunx skills add JiayiV\u002Fquestion_mixing --skill inbox-triage",{"tool":71,"command":72},"pnpm","pnpm dlx skills add JiayiV\u002Fquestion_mixing --skill inbox-triage","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fpublic-02-claude-skills-main-gemini-skills-inbox-triage-skill-e510c8d8\u002FSKILL.md",true,[],[77,103,129,160,188,223],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":46},"01M6NN9PJS9NHWPDG4QP93BP8F","webmcpify","Make a web app agent-ready — propose a WebMCP tool manifest, integrate, verify in a real browser, heal; unrelated code stays untouched. Use for \"webmcpify\", \"add WebMCP\", or \"expose app actions to AI agents\".","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fwebmcpify\u002FSKILL.md",[],[79,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102],"github","ai agents","web","notes","code","browser","https","actions","manifest","inventory","verify","tool","use","ready","unrelated","app","scope","for","webmcp",{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":107,"licenseFileUrl":6,"sourceHtmlUrl":108,"categories":109,"keywords":110,"publishedAt":46},"01M6NN6KZE5RBN6NGHN2WXV1RE","vardoger-analyze","Use when the user asks to personalize the GitHub Copilot CLI assistant, adapt Copilot to their style, use vardoger, or analyze their Copilot CLI conversation history. Reads the local session directory at `~\u002F.copilot\u002Fsession-state\u002F`, extracts recurring preferences and conventions, and writes a fenced personalization block into `~\u002F.copilot\u002Fcopilot-instructions.md`. Runs entirely on the user's machine via the local `vardoger` CLI (`pipx install vardoger`); no network calls and no uploads. Triggers: 'personalize my copilot', 'analyze my copilot history', 'tailor copilot to me', 'run vardoger', 'update my copilot instructions from history', 'make copilot learn my style'.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fvardoger-analyze\u002FSKILL.md",[],[111,112,113,114,115,116,45,117,118,119,120,121,122,34,123,124,125,126,127,128],"vardoger analyze","analyze","directory","cli","github copilot","copilot","update","learn","network","my style","install","conversation","local","github copilot cli","pipx","my","at","copilot instructions",{"skillNo":130,"name":131,"nameZh":6,"description":132,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":133,"categories":134,"keywords":141,"publishedAt":46},"01M6NN3CZGQP3XAJ04BM1NNYGT","transloadit-media-processing","Process media files (video, audio, images, documents) using Transloadit. Use when asked to encode video to HLS\u002FMP4, generate thumbnails, resize or watermark images, extract audio, concatenate clips, add subtitles, OCR documents, or run any media processing pipeline. Covers 86+ processing robots for file transformation at scale.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Ftransloadit-media-processing\u002FSKILL.md",[135,138],{"categoryNo":136,"categoryName":137,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000016","图像与多媒体",{"categoryNo":139,"categoryName":140,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000023","Skill与智能体开发",[142,143,144,145,146,147,148,149,150,151,152,153,154,96,155,156,157,158,101,159],"transloadit media processing","Media","media processing","processing","MCP","mcp server","documents","subtitles","pipeline","ocr","Video","Audio","mp4","images","files","transformation","free","hls",{"skillNo":161,"name":162,"nameZh":6,"description":163,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":164,"categories":165,"keywords":169,"publishedAt":46},"01M6NN28E92F9GH4WFYQ2H1Y15","terraform-azurerm-set-diff-analyzer","Analyze Terraform plan JSON output for AzureRM Provider to distinguish between false-positive diffs (order-only changes in Set-type attributes) and actual resource changes. Use when reviewing terraform plan output for Azure resources like Application Gateway, Load Balancer, Firewall, Front Door, NSG, and other resources with Set-type attributes that cause spurious diffs due to internal ordering changes.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fterraform-azurerm-set-diff-analyzer\u002FSKILL.md",[166],{"categoryNo":167,"categoryName":168,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000011","运维与部署",[170,171,172,173,174,175,176,177,178,96,179,180,181,182,183,184,185,186,187,101],"terraform","SET","terraform azurerm set diff analyzer","diff","analyzer","diffs","azure","gateway","firewall","only","resource changes","provider","resources","load balancer","attributes","front","load","balancer",{"skillNo":189,"name":190,"nameZh":6,"description":191,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":192,"categories":193,"keywords":205,"publishedAt":46},"01M6NN1QMH2TDYWTH3XY3N5XPV","technical-job-search","Use this skill when a software engineer asks for help with job search tasks: parsing or analyzing a job description, tailoring a CV\u002Fresume, writing a cover letter, evaluating a job offer, or drafting a post-interview follow-up email. Do not activate for general career advice unrelated to an active job search action.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Ftechnical-job-search\u002FSKILL.md",[194,197,202],{"categoryNo":195,"categoryName":196,"taxonomyType":20,"taxonomyTypeDesc":21},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":198,"categoryName":199,"taxonomyType":200,"taxonomyTypeDesc":201},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",{"categoryNo":203,"categoryName":204,"taxonomyType":200,"taxonomyTypeDesc":201},"01JBSEEDCAT000000000000038","人力资源",[206,207,208,209,210,211,212,213,112,214,96,215,216,217,218,219,220,221,101,222],"search","Job Search","technical job search","technical","writing","tasks","email","parsing","interview","up","engineer","general","activate","do","task","cv","letter",{"skillNo":224,"name":225,"nameZh":6,"description":226,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":227,"categories":228,"keywords":229,"publishedAt":46},"01M6NMZFA67XYS1GS09WT42235","steno-mode","Shorthand-first response compression that cuts ~40% of response tokens while preserving technical precision and exact literals. Use when the user says \"steno mode\", \"shorthand mode\", \"compressed responses\", \"token reduction\", \"brief structured output\", or invokes \u002Fsteno. Supports four compression levels: lite, brief, court, machine. Do not trigger for requests needing polished prose such as onboarding\u002Ftutorial content, stakeholder or customer-facing copy, or teaching-focused explanations.","https:\u002F\u002Fgithub.com\u002FJiayiV\u002Fquestion_mixing\u002Fblob\u002Fmain\u002Fskills\u002Fsteno-mode\u002FSKILL.md",[],[230,231,232,233,234,209,235,96,219,236,237,238,239,240,241,242,243,244,101,245],"steno mode","mode","customer","brief","first","tutorial","teaching","tokens","as","trigger","structured output","machine","output","court","requests","response compression"]