Linkt Outreach Skill

Draft personalized LinkedIn connection requests and post to Slack for manual sending.

Configuration

Slack Channel: #linkt-connections (ID: C0ACE4HJLBD)

Prerequisites

This skill requires Slack MCP to be configured. See docs/slack-setup.md for setup instructions.

Input Context

This skill expects one of:

1. •From /linkt-signals: Signal context, company info, and contact with LinkedIn URL
2. •Direct invocation: User provides LinkedIn URL and optional context

Required information:

• •Contact LinkedIn URL
• •Contact name and title
• •Company name

Optional but helpful:

• •Signal summary (what triggered the outreach)
• •Company context (industry, what they do)
• •Shared interests or relevant talking points

Workflow

Step 0: Load User Context

First, try to read .claude/user-context.json for personalization context.

If the file exists, load the user context for use in message drafting:

• •Company name and description
• •User's role
• •Value proposition
• •Talking points
• •User's name (for sign-off)

If the file doesn't exist, continue without user context. Messages will be more generic but still functional. Optionally mention: "Tip: Run /linkt-init to set up your profile for more personalized messages."

Step 1: Gather Context

If invoked directly (not from /linkt-signals), ask the user for:

• •LinkedIn profile URL
• •Why they want to connect (signal, shared interest, etc.)
• •Any specific talking points

If context was passed from /linkt-signals, confirm the details:

**Outreach Context:**
- Contact: [Name] ([Title])
- Company: [Company Name]
- Signal: [Signal summary]
- LinkedIn: [URL]

Is this correct? (yes/edit)

Step 2: Draft Connection Message

Craft a personalized LinkedIn connection request (max 200 characters for connection note).

Message Guidelines:

• •Lead with relevance (the signal or shared interest)
• •Be specific about why you're reaching out
• •Keep it concise and professional
• •Avoid generic templates
• •Don't be salesy in the connection request

Personalization Rules (when user context is available):

If user context was loaded from .claude/user-context.json:

1. •

   Subtle company reference: Mention user's company naturally when relevant to the signal. Don't force it.

   • •Good: "At [Company], we help teams like yours with [relevant capability]"
   • •Bad: "[Company] is the #1 solution for..." (too salesy)
2. •

   Value proposition: Weave in the value proposition when it connects to the signal context.

   • •If the signal is about AI initiatives and user helps with AI, mention the connection
   • •If there's no natural fit, omit it entirely
3. •

   Talking points: Use a talking point only if directly relevant to the contact's situation.

   • •Recent funding → mention if discussing growth/scaling
   • •Press coverage → mention if it relates to the signal topic
   • •Don't shoehorn talking points where they don't fit
4. •

   Sign-off with name: If user.name is set, end with "- [Name]" for a personal touch.

5. •

   Tone alignment: Match the use_case:

   • •sales → professional but warm, focus on mutual value
   • •recruiting → friendly and opportunity-focused
   • •partnerships → collaborative and strategic
   • •networking → casual and interest-based

Message Template Structure:

Hi [First Name],

[1-2 sentences referencing the signal and why connecting would be valuable]

[Optional: sign-off with user's name if available]

Example Messages:

Without user context (generic):

Hi Sarah,

Your recent article on AI transformation resonated with me. Would love to connect and exchange insights.

With user context (personalized):

Hi Sarah,

Your article on AI in sales caught my attention - at Linkt, we help GTM teams discover prospects based on these signals. Would love to connect!

- Jack

Step 3: Present Draft for Approval

Show the drafted message to the user:

## LinkedIn Connection Request

**To:** [Name] ([Title]) at [Company]
**Profile:** [LinkedIn URL]

**Draft Message:**
---
[Message content]
---

Character count: [X]/200

**Actions:**
- Type 'send' to post to Slack
- Type 'edit' to modify the message
- Type 'cancel' to abort

Step 4: Handle User Response

If 'edit': Ask the user what they'd like to change, then redraft and present again.

If 'cancel': Acknowledge and exit gracefully.

If 'send': Proceed to Slack posting (Step 5).

Step 5: Post to Slack for Manual Execution

After user approves the draft, post to #linkt-connections (channel ID: C0ACE4HJLBD).

Use mcp__slack__slack_post_message with channel_id: "C0ACE4HJLBD".

Message format:

📬 *LinkedIn Connection Request*

*To:* [Name] ([Title]) at [Company]
*Profile:* [LinkedIn URL]
*Signal:* [Signal summary]

*Draft Message (200 char max):*
> [Message content]

_Copy the message above and send via LinkedIn._

Step 6: Confirmation

After posting to Slack:

**Posted to Slack!**

- Channel: #linkt-connections
- Contact: [Name] at [Company]
- Signal: [Signal type]

**Next steps:**
1. Open the LinkedIn profile: [URL]
2. Click "Connect" → "Add a note"
3. Paste the message from Slack
4. Send the connection request

**Tip:** Follow up in 3-5 days if they accept the connection.

Safety Features

1. •Always show draft before posting - User must approve the message
2. •Human executes the action - Claude posts to Slack, human sends on LinkedIn
3. •Persistent record - Slack provides audit trail of outreach
4. •No automation complexity - Avoids browser automation issues

Error Handling

• •Slack not configured: Provide manual instructions and suggest running docs/slack-setup.md
• •LinkedIn URL invalid: Ask user to verify the URL
• •Channel not found: List available channels and ask user to choose

Character Limits

• •LinkedIn connection note: 200 characters max
• •Keep messages under 180 characters to be safe
• •If message is too long, offer to trim it

Do NOT

• •Post to Slack without user approval of the draft
• •Use generic/template messages without personalization
• •Mention specific product pitches in connection requests
• •Skip the confirmation step