Agent skill
slack-writing
Generates Slack messages in user's authentic voice: direct, concise, data-driven. Use when user needs to write, draft, compose, or rewrite Slack messages, DMs, or channel posts. Adapts style based on channel type (public vs DM) and message type.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/slack-writer
SKILL.md
Slack Message Writer Skill
Purpose
Generate Slack messages in your authentic voice: direct, concise, data-driven, and confidently human. Adapts style based on channel type (public vs DM) and message type (status, decision, concern, etc.).
Voice Profile
Core Characteristics
- Direct and concise: Short punchy sentences. No fluff.
- Data-driven: Specific metrics presented in bullet format with context.
- Confident: Clear opinions stated directly. Explicit confidence levels when uncertain.
- Action-oriented: Every message ends with clear next steps and deadlines.
- Structured: Bullets and line breaks for organization.
- Professional but human: Occasional emojis for emphasis (✅ 🔴 📈), minimal contractions.
Style Rules
- Short, punchy sentences - Get to the point fast
- No hedging words - Avoid "maybe", "possibly", "I think perhaps", "it seems like"
- Confidence levels for uncertainty - "Medium confidence here" / "High confidence this works"
- Bullet format for data/lists - Always contextualize metrics
- Action-oriented closings - Include deadlines and owners
- Occasional emojis - ✅ 🔴 📈 for emphasis only
- Minimal contractions - Prefer "it is" over "it's", "we are" over "we're"
- Pronoun usage - "I" for opinions, "we" for team actions
- @mentions for urgency - Not "URGENT:" labels
Channel Type Detection
Public Channels (team-wide, project channels)
- Opening: Use context setter - "Quick update:" / "FYI:" / "Heads up:"
- Tone: Same direct style, slightly more structured
- Example opener: "Quick update: shipped the landing page redesign yesterday."
DMs or Small Groups
- Opening: Jump straight to content - no prefix needed
- Tone: Same direct style
- Example opener: "Shipped the landing page redesign yesterday."
Thread Responses
- Opening: Quote or reference original message
- Pattern: "You mentioned X - here is my take..." or "On the API decision:"
- Tone: Conversational but still direct
Message Type Templates
1. Status Update
Quick update: [what happened in one line].
[Metrics if applicable:]
• [Metric 1]: [Value] ([context - above/below target])
• [Metric 2]: [Value]
Next steps:
• [Action 1] - [Owner] - [Deadline]
• [Action 2] - [Owner] - [Deadline]
Example:
Quick update: shipped landing page redesign yesterday.
Early metrics:
• Adoption: 15% (below 20% target)
• Page load: 1.2s (within goal)
Next steps:
• @sarah run user interviews by Friday
• @mike pull funnel analytics by EOD Thursday
Will synthesize findings and share action plan Monday.
2. Asking for Input/Decision
Need input on [topic].
Context: [1-2 sentences of essential background]
Options:
• Option 1: [Brief description] - [Trade-off]
• Option 2: [Brief description] - [Trade-off]
[Your position]: Leaning toward [option] because [reason].
@[person] need your call by [deadline].
Example:
Need input on API approach.
Context: Building the integration layer. Two viable paths with different trade-offs.
Options:
• Option 1: REST - faster to ship (2 weeks), creates tech debt
• Option 2: GraphQL - cleaner long-term, delays launch by 2 weeks
My take: Leaning toward Option 1 given our Q1 timeline pressure.
@david need your call by EOD Wednesday.
3. Raising Concern/Risk
Flagging a risk: [one-line summary].
Context: [Why this matters]
Impact: [What happens if not addressed]
Recommendation: [Proposed action]
[Confidence level if uncertain]: Medium confidence this is the right call.
Example:
Flagging a risk: Legal approval timeline is tight.
Context: Current plan assumes 5-day Legal review. Based on past reviews, that is optimistic.
Impact: Can slip launch by 2+ weeks if not addressed now.
Recommendation: Either add 2-week buffer or start Legal review this week.
High confidence this needs attention.
4. Disagreeing/Alternative View
Different take: [your position in one line].
[Why you see it differently - 2-3 sentences max]
Trade-off to consider: [What your approach gains/loses]
Open to discussing if helpful.
Example:
Different take: we should prioritize mobile over desktop.
Usage data shows 65% of our users are on mobile during site visits. Desktop-first means we are optimizing for the minority. The mobile experience is also where we are losing users in the funnel.
Trade-off: Desktop gets delayed by 1 sprint, but we capture the bigger audience first.
Open to discussing if helpful.
5. Positive Feedback/Celebrating Wins
[Impact statement]. Well done.
What this unlocks: [Customer/business value]
[Specific callout if applicable]: @[person] [specific contribution]
Example:
This unlocks real-time collaboration for our largest customer segment. Well done.
What this unlocks: Enterprise teams can now work simultaneously without sync delays.
@lisa your work on the conflict resolution logic made this possible.
6. Follow-up/Check-in
Checking in on [topic].
Last status: [What was agreed/expected]
Need by: [Deadline]
Let me know if anything is blocking progress.
Example:
Checking in on the API documentation.
Last status: Draft was due yesterday per our sync.
Need by: Friday to unblock partner integration testing.
Let me know if anything is blocking progress.
7. Announcement
FYI: [What is happening in one line].
Key details:
• [Detail 1]
• [Detail 2]
• [Detail 3]
What this means for you: [Specific impact/action needed]
Questions? Thread here or DM me.
Example:
FYI: Launching the new dashboard to all users next Monday.
Key details:
• Rollout: 10% Monday, 50% Wednesday, 100% Friday
• Training: Self-serve guide in Help Center
• Support: #dashboard-questions for issues
What this means for you: Review the guide before Monday. Flag concerns by Friday.
Questions? Thread here or DM me.
Thread Response Patterns
Agreeing and Adding
+1 on this. Adding: [your contribution]
Providing Requested Input
You asked about [topic]. Here is my take:
[Your input in bullet format if multiple points]
Clarifying a Question
On your question about [topic]:
[Direct answer]
[Context if needed - 1 sentence max]
Anti-Patterns to Avoid
- ❌ Hedge words: "I think maybe we could possibly consider..."
- ❌ Over-explaining: Long paragraphs of context before the point
- ❌ Burying the ask: Action item hidden at the end
- ❌ Vague deadlines: "soon" / "when you get a chance" / "ASAP"
- ❌ Excessive emojis: 🎉🚀💪🔥 everywhere
- ❌ URGENT labels: Use @mentions instead
- ❌ Multiple unrelated topics: One message, one topic
- ❌ Passive voice: "It was decided that..." instead of "We decided..."
- ❌ Unnecessary greetings: "Hey! Hope you are having a great day!"
Quality Checklist
Before sending, verify:
- Point is clear in first line
- Action/ask is explicit with owner and deadline
- Data has context (vs target, vs last period, etc.)
- No hedge words or passive voice
- Length is appropriate (shorter is better)
- Emojis used sparingly for emphasis only
- Would a PM leader approve this level of clarity?
Confidence Level Expressions
Use these when uncertain:
- High confidence: "High confidence this is the right call."
- Medium confidence: "Medium confidence here. Worth validating with [person/data]."
- Low confidence: "Low confidence - flagging for discussion, not recommending action yet."
Organizational Context
For product-specific context (product name, key terms, stakeholders), see CLAUDE.local.md.
When drafting Slack messages:
- Stakeholders: Reference
Reference/corporate-strategy/for org context - Initiatives: Reference
Initiatives/*/CLAUDE.mdfor project context - Colleagues: Reference
Reference/colleagues.jsonfor name verification
Didn't find tool you were looking for?