linkedin-message-writer

# LinkedIn Message Writer

Research LinkedIn leads and write personalized messages for any LinkedIn message type. Takes LinkedIn URLs, researches each person using Apify (profile + recent posts), and writes messages based on what it finds.

No LinkedIn cookies. No database setup. Just LinkedIn URLs in, personalized messages out.

## When to Auto-Load

Load this skill when:
- User says "write LinkedIn messages", "LinkedIn outreach", "connect with these leads on LinkedIn", "send LinkedIn messages"
- User has a list of LinkedIn URLs and wants to reach out
- User wants to write personalized connection requests, InMails, DMs, or comments

## Prerequisites

### Apify API Token

Required for researching LinkedIn profiles and posts. Set in `.env`:

```
APIFY_API_TOKEN=your_token_here
```

No LinkedIn cookies, login, or session tokens needed. Apify handles scraping without any LinkedIn credentials.

That's it. One env var. Nothing else.

---

## LinkedIn Message Types Reference

This skill writes any text-based LinkedIn message type. Each type has different constraints.

| Message Type | Who Can Receive | Character Limit | When to Use |
|-------------|----------------|-----------------|-------------|
| **Connection request** | 2nd/3rd degree connections | 200 (free) / 300 (premium) | First touch. Must earn the accept. No selling. |
| **InMail** | Anyone (requires premium credits) | Subject: 200, Body: 1,900 | Standalone pitch to people who won't accept cold connections. Senior execs, busy people. |
| **DM** | 1st-degree connections only | 8,000 | Follow-ups after connection accepted. Conversational, not broadcast. |
| **Message request** | Group members, event attendees, #OpenToWork | 8,000 | Warm context — you share a group or event. Reference the shared context. |
| **Post comment** | Anyone (public posts) | 1,250 | Warm-up before connecting. Show you engaged with their content. Not a pitch. |
| **Comment reply** | Anyone (in a thread) | 1,250 | Engage in a conversation they started. Add value, don't pitch. |

### Key Rules Per Type

**Connection request (200/300 chars):**
- This is the gatekeeper. If they don't accept, nothing else happens.
- Lead with the signal — what they did/said/posted that caught your attention.
- One sentence of relevance. No pitch, no CTA, no "I'd love to..."
- MUST be under the character limit. Count every character. If over, rewrite — never truncate.
- Free accounts: 200 chars. Premium/Sales Navigator: 300 chars. Ask the user which they have.

**InMail (subject 200 + body 1,900 chars):**
- Must work standalone — they haven't accepted your connection.
- Subject: curiosity-driven, not salesy. Not "Quick question" or "Partnership opportunity."
- Body: include context for why you're reaching out (the signal). Be specific.
- Higher commitment ask is OK here — you're using a premium credit.

**DM (8,000 chars):**
- Conversational. These read like DMs, not emails.
- Shorter is almost always better. A 2-sentence message outperforms a 5-sentence one.
- Good for follow-up sequences after connection accepted.
- Sequence structure: Day 0 connection → Day 3 value-first → Day 7 social proof → Day 14 breakup.

**Message request (8,000 chars):**
- Always reference the shared context (group name, event name, OpenToWork status).
- More casual than InMail since you have something in common.

**Post comment (1,250 chars):**
- Add genuine value. Share an insight, ask a smart question, build on their point.
- NOT "Great post!" or "Love this!" — that's noise.
- This is a warm-up move, not a pitch. The goal is to get noticed before connecting.

**Comment reply (1,250 chars):**
- Continue the conversation. Reference what they said specifically.
- Shorter than a standalone comment. 2-3 sentences max.

---

## Workflow

### Phase 0: Intake

Ask the user these questions. Skip any already answered.

**Leads:**
1. Where are your leads? (CSV file, paste LinkedIn URLs, database, CRM — whatever they have)
2. How many leads? (affects cost estimate and whether to use post scraper)

**Message type:**
3. What kind of LinkedIn message do you want to write? (connection request, InMail, DM, message request, post comment, comment reply, or a sequence of multiple types)
4. If connection request: do you have a free or premium LinkedIn account? (affects character limit: 200 vs 300)

**Goal:**
5. What's the objective? (book meetings, drive demo requests, get replies, build relationships, promote content, warm up before outreach)
6. What's the angle or hook? (pain-based, hiring signal, competitor displacement, event-based, content engagement, mutual connection, cold)

**Tone:**
7. Which tone? Present options:
   - **Casual Professional** — Friendly, human, slightly informal. Like messaging a peer. (default)
   - **Thought Leader** — Lead with insight or a contrarian take. Position sender as expert.
   - **Provocative** — Challenge assumptions, pattern-interrupt. Higher risk, higher reward.
   - **Enterprise Formal** — Polished, structured. For regulated industries or C-suite targets.
   - **Custom** — User pastes reference messages that have worked, or describes the vibe.
8. Any reference messages that have worked well? (these override tone presets)

**Context:**
9. What does your company/product do? (one-liner for the AI to work with)
10. Any proof points? (customer names, metrics, case studies to reference)

**Output:**
11. Which LinkedIn outreach tool do you use? (Dripify / Expandi / Botdog / PhantomBuster / Just give me a CSV)

### Phase 1: Load Leads

Accept leads from whatever source the user provides:

- **CSV file:** Read the CSV. Look for a column containing LinkedIn URLs (common names: `linkedin_url`, `LinkedIn URL`, `LinkedIn`, `profile_url`, `url`). If ambiguous, ask the user which column.
- **Pasted URLs:** User pastes LinkedIn URLs directly. Parse them.
- **Pasted list:** User pastes names + companies or other data. Extract what's available.
- **Database/CRM:** Ask the user how to access it. Use w