skills/email-composer/SKILL.md
Use when composing a NEW email from scratch (any type: business, technical, customer, personal-professional). NEVER for editing/polishing existing drafts (email-draft-polish), multi-email sequences or drip campaigns (email-sequence), email infrastructure/deliverability/SPF/DKIM (email-systems), cold sales outreach (cold-email), internal company announcements/memos (internal-comms), general text editing (copy-editing), or non-email professional writing (professional-communication).
npx skillsauth add sharkitect-solutions/sharkitect-claude-toolkit Email ComposerInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
| File | Purpose | Load When | |------|---------|-----------| | SKILL.md | Tone calibration matrix, subject line engineering, structural anti-patterns (8), recipient psychology (4 types + cross-cultural), email failure checklist | Always (auto-loaded) | | scenario-frameworks.md | Structural frameworks for 8 email scenarios: meeting request, follow-up (5 types), introduction (3 types), feedback request, status update, decline/say-no, decision request, apology, vendor negotiation | When composing an email for a specific scenario that needs structural guidance | | difficult-emails-guide.md | Bad news delivery (3 severity levels + medium decision), negative feedback by email (when appropriate vs dangerous), escalation emails (pre-checklist + structure), declining without burning bridges, expectation management, conflict de-escalation (5-step + escalating vs de-escalating words), legal-sensitive communications | When the email involves sensitive, high-stakes, or emotionally charged content | | email-timing-and-follow-up.md | Optimal send times (6 recipient types), follow-up timing and cadence (5 urgency levels), timezone management (5 scenarios), email vs other medium decision tree (8 message types), email chain management (6 situations), read receipt ethics, inbox management for recipients | When timing matters, planning follow-ups, or deciding whether email is the right medium |
| Need | Use This Skill? | Use Instead | |---|---|---| | Compose a new email from scratch | YES | -- | | Edit/polish an existing draft | NO | email-draft-polish | | Multi-email sequences or drip campaigns | NO | email-sequence | | Email infrastructure, deliverability, SPF/DKIM | NO | email-systems | | Cold sales outreach to strangers | NO | cold-email | | Internal memos, announcements, org comms | NO | internal-comms | | General text editing (not email) | NO | copy-editing | | Non-email professional writing | NO | professional-communication |
Select tone parameters by crossing recipient relationship with purpose. Do NOT default to "professional" -- calibrate precisely.
| Recipient | Request | Update/FYI | Bad News | Escalation | Appreciation | |---|---|---|---|---|---| | C-suite | BLUF, 0 contractions, no exclamation, "Regards" | 3 sentences max, bullet data, "Best" | Lead with impact number, then cause, then fix, "Regards" | State risk in dollars/days, 1 clear ask, "Respectfully" | 1 sentence specific praise, tie to metric, "Best" | | Direct manager | Light contractions OK, 1 exclamation max, "Thanks" | Bullet-heavy, include your interpretation, "Best" | Own it first sentence, corrective action second, "Thanks" | Frame as blocker to shared goal, propose options, "Thanks" | Genuine, specific, skip superlatives, "Thanks" | | Peer/colleague | Contractions normal, casual sign-off, "Cheers"/"Thanks" | Conversational, skip preamble, "Thanks" | Direct but empathetic, offer help, "Thanks" | Name the impact on your work, no blame, "Thanks" | Informal, can use humor if relationship supports it, "Cheers" | | Direct report | Warm but clear, contractions OK, "Thanks" | Add context on WHY, not just what, "Best" | Private, specific, forward-looking, "Thanks" | Be direct about gap, ask for their plan, "Thanks" | Public when possible, name specific behavior, "Great work" | | External client | 0 contractions, 0 exclamations, "Kind regards" | Lead with value to THEM, "Best regards" | Acknowledge impact first, solution second, timeline third, "Sincerely" | Restate SLA/agreement, document facts, "Regards" | Tie to business outcome, "Best regards" | | Vendor/partner | Direct, transactional, "Best" | Brief, action-oriented, "Thanks" | State terms, cite agreement, "Regards" | Quote contract clause, escalation path, "Regards" | Brief acknowledgment, "Thanks" |
Sentence length: C-suite/client = 12-18 words avg. Peers = 8-15 words. Technical audiences = whatever clarity requires. Emoji policy: Never in first email to someone. Never to C-suite/clients. Peers only after they use emoji first.
Mobile preview truncates at 30 characters. Front-load the essential word.
| Pattern | When to Use | Example |
|---|---|---|
| [ACTION REQUIRED] prefix | Recipient must do something by a date | [ACTION REQUIRED] Approve budget by Fri |
| [FYI] prefix | No action needed, awareness only | [FYI] API latency spike resolved |
| [DECISION NEEDED] prefix | Binary/ternary choice required | [DECISION NEEDED] Vendor A or B for Q3 |
| Naked subject (no prefix) | Standard correspondence | Database migration timeline update |
| Re: chain continuation | Same thread, same topic | Keep existing subject |
| Break Re: chain when | Topic shifted, new decision needed, or chain > 5 deep | New subject referencing old thread |
Internal vs external: Internal subjects can use prefixes and abbreviations. External subjects to clients must read as complete, professional phrases -- no brackets, no abbreviations.
Never: Single-word subjects ("Update", "Question", "Hello"). Clickbait urgency when not urgent. ALL CAPS words.
These patterns cause emails to fail (no response, wrong action, confusion). Detect and fix before output.
1. Buried Lede -- Action item appears in paragraph 3+. Fix: First sentence = what you need. Second sentence = by when. Context follows.
2. Essay Email -- >150 words for a request that needs <40. Fix: If the ask fits in 2 sentences, the email is 2 sentences + sign-off. Permission to be brief.
3. Passive Ask -- "It would be great if someone could look into this." Fix: Name the person. Name the action. Name the date. "Alex, please review the PR by Thursday EOD."
4. Premature Apology -- "Sorry to bother you" / "I know you're busy" as openers. Fix: Delete. These phrases signal low status and prime the reader to deprioritize. Just make the ask.
5. Wall of Context -- 5 paragraphs of background before revealing the ask. Fix: BLUF (Bottom Line Up Front). Ask first, context second, detail in attachment if needed.
6. Multi-Ask Sprawl -- 3+ unrelated requests in one email. Fix: One email = one ask. Completion rate drops ~50% per additional ask. Send separate emails or use numbered list with explicit "I need responses to all 3."
7. Ambiguous Deadline -- "When you get a chance" / "at your earliest convenience." Fix: Specific date and time. "By Wednesday 3 PM ET" or "No rush -- next week is fine" (at least anchors expectation).
8. Attachment Phantom -- "See attached" but no attachment; or attachment not referenced in body. Fix: Always verify attachment is present. Reference it by name: "See the attached Q4-forecast.xlsx for details."
Structure: BLUF, then 3 bullets max, then "Happy to provide detail." Attach the detail doc -- never paste it inline. They scan, they don't read. Make scanning productive. Every sentence must either state the ask, quantify impact, or propose a decision.
Skip pleasantries. Lead with the technical context they need to evaluate the request. Include version numbers, error codes, links to logs. "Hi, hope you're well" wastes their time and yours. Bullet points > paragraphs. Code blocks welcome.
First sentence acknowledges their world ("I know launch week is intense"). Quantify value to THEM, not to you. Every email ends with an explicit next step that includes a date: "I'll send the revised mockups by Thursday. If I don't hear back by Friday, I'll proceed with version B." Never leave next steps ambiguous.
Be transactional. State what you need, when, and the terms. Document agreements in email even if discussed on a call ("Per our call today, confirming: ..."). Friendly but not warm -- warmth gets exploited in negotiations.
| Culture | Directness | Formality | Preamble | "No" Style | |---|---|---|---|---| | US | Direct ask OK | Medium | Brief | "Unfortunately, we can't" | | UK | Slightly indirect | Medium-high | Polite buffer | "I'm afraid that won't be possible" | | Japan | Indirect, consensus-seeking | High | Seasonal greeting expected | Imply difficulty, never blunt refusal | | Germany | Very direct, expected | High initially | Minimal | Direct "no" is professional, not rude | | Latin America | Relationship-first | Medium | Personal warmth expected | Soft decline with alternative |
Before outputting any composed email, verify against these failure modes:
| Temptation | Why It Fails | Do Instead | |---|---|---| | "I'll add context so they understand fully" | Reader skips long emails. Understanding != reading. | BLUF + "Details below if needed" with a visual break | | "I should soften this with hedging language" | "Maybe we could possibly consider" = no one acts | State the ask directly. Politeness is in tone, not hedging. | | "I'll CC the whole team for visibility" | Diffusion of responsibility. Everyone assumes someone else will act. | TO: the person who must act. CC: only those who must know. | | "The formal template will look more professional" | Generic templates read as generic. | Match the relationship. Forced formality to a peer feels cold. | | "I'll include all the background they might need" | Information overload kills action. | Include only what's needed to decide. Link or attach the rest. | | "I should follow up same-day to show urgency" | Reads as impatient or aggressive. | Wait 2-3 business days. Then follow up with new value, not "just checking in." |
[bracketed placeholders] -- ask for the missing information, then compose a complete emaildevelopment
When the user wants help with paid advertising campaigns on Google Ads, Meta (Facebook/Instagram), LinkedIn, Twitter/X, or other ad platforms. Also use when the user mentions 'PPC,' 'paid media,' 'ad copy,' 'ad creative,' 'ROAS,' 'CPA,' 'ad campaign,' 'retargeting,' or 'audience targeting.' This skill covers campaign strategy, ad creation, audience targeting, and optimization.
testing
--- name: using-sharkitect-methodology description: Use when starting any conversation in a Sharkitect workspace OR before any task involving NEW pricing, positioning, proposal, strategy, plan-execution, or schema-design work — mandates invocation of Sharkitect-specific methodology skills (pricing-strategy, marketing-strategy-pmm, smb-cfo, hq-revenue-ops, executing-plans, brainstorming) under the same anti-rationalization discipline as using-superpowers. Documentation has failed 4 times across H
testing
Use when user says 'end session', 'wrap up', 'stop for the day', 'done for today', 'close out', 'save session', 'wrapping up', or invokes /end-session. Runs the full 9-step end-of-session protocol: resource audit, MEMORY.md update, lessons capture, plan status, pending items, workspace checklist, .tmp/ audit, git commit+push, Supabase brain sync, session brief, summary. Final step schedules a detached self-kill of the current session ONLY (3s delay) so the window closes cleanly. Other claude.exe processes (active workspaces) are NOT touched -- orphan cleanup is handled separately by Claude-Orphan-Cleanup-Hourly with proper age safeguards. Do NOT use for: mid-session quick saves (use session-checkpoint), skill syncing (use sync-skills.py), brain memory queries (use supabase-sync.py pull), document freshness reviews (use document-lifecycle), resource gap detection (use resource-auditor).
testing
Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, passive voice, negative parallelisms, and filler phrases.