coffee-chat

What this skill does

# Coffee Chat Playbook Generator

This skill helps you prepare for professional coffee chats by gathering intelligence on the person and generating personalized conversation guides.

---

## Prerequisites

Before using this skill, set up the following. Only **Apify CLI** is required for X scraping — everything else is optional.

### 1. Apify CLI (required for X/Twitter scraping)

Apify CLI is used to run the tweet scraper actor.

**Install:**
```bash
npm install -g apify-cli
```

**Authenticate:**
```bash
apify login
# Paste your Apify API token when prompted
# Get your token at: https://console.apify.com/account/integrations
```

**Verify:**
```bash
apify info
# Should show your username and token
```

> Free tier includes $5/month of compute credits. Scraping 10 tweets costs ~$0.004, well within free limits.

---

### 2. Notion Integration (optional — for saving playbooks to Notion)

**Create an integration:**
1. Go to https://www.notion.so/my-integrations
2. Click **"New integration"**
3. Give it a name (e.g. "Coffee Chat Skill")
4. Copy the **Internal Integration Token**

**Set the API key:**
```bash
export NOTION_API_KEY="secret_xxxxxxxxxxxxxxxxxxxx"
# Add to your shell profile to persist:
echo 'export NOTION_API_KEY="secret_xxxxxxxxxxxxxxxxxxxx"' >> ~/.zshrc
```

**Share your Notion page with the integration:**
1. Open the Notion page where you want playbooks saved
2. Click **"..."** → **"Add connections"** → select your integration
3. Copy the page ID from the URL:
   - URL: `https://notion.so/My-Page-abc123def456...`
   - Page ID: `abc123def456...` (the part after the last `-`)

**Update `NOTION_PAGE_ID`** in the Notion Push section of this skill.

---

### 3. Web Search & Fetch (required for company and background research)

This skill relies on web search and web fetch to research the target's company, industry, and background in Steps 2 & 3. Your AI agent needs access to these tools.

**If using OpenClaw or Claude Code:**
- Web search and WebFetch are built-in — no setup needed

**If using another agent framework**, make sure it has access to:
- A **search API** (e.g. Brave Search, Serper, Tavily, You.com) for querying company news, funding rounds, founder backgrounds
- A **web fetch / scrape tool** for reading pages (company websites, LinkedIn profiles, news articles)

> Without these tools the agent will fall back to training knowledge only, which may be outdated. Research quality will drop significantly.

---

### 4. Summary of everything you need

| Tool / Variable | Required | Description |
|-----------------|----------|-------------|
| Web search API | **Yes** (for research) | Built-in on OpenClaw/Claude Code; otherwise configure your agent's search plugin |
| Web fetch | **Yes** (for research) | Built-in on OpenClaw/Claude Code; otherwise configure your agent's fetch tool |
| `APIFY_API_TOKEN` | For X scraping | Set automatically by `apify login` |
| `NOTION_API_KEY` | Optional | Your Notion integration token |

---

## Setup (First Time)

Before using this skill, create your profile file at `memory/my-profile.md`:

```markdown
# My Profile

- **Name**: [Your Name]
- **Role**: [Your current role and company]
- **Background**: [Brief career background, e.g. Finance → VC → AI]
- **Location**: [City, Country]
- **LinkedIn**: [Your LinkedIn URL]
- **Interests**: [Topics you care about / want to discuss]
```

Also configure integrations (optional):
- **Notion**: Set `NOTION_API_KEY` env var and update `NOTION_PAGE_ID` in the Notion Push section below
- **Apify**: Set `APIFY_API_TOKEN` env var for LinkedIn scraping (free tier available)

---

## Workflow

### Step 1: Collect Information

Ask the user for:
1. **Target person's name** (required)
2. **Target's LinkedIn profile URL** (optional but recommended)
3. **Target's company name** (required for research)
4. **Target's X (Twitter) username** (optional, e.g. `@elonmusk` or just `elonmusk`) — used for scraping recent posts
5. **Context of the meeting** (optional - how you met, goal)

Load your profile from `memory/my-profile.md` for comparison.

### Step 2: Research Target's Company

For each target, do comprehensive company research:

1. **Web search** for company info:
   - Funding news
   - Product/feature launches
   - Team / founders

2. **X (Twitter) search** for company account:
   - Recent posts
   - Product demos
   - Engagement with community

3. **News search** for recent coverage:
   - Funding announcements
   - Product launches
   - Competitor/industry analysis

### Step 3: Research Target's Background

1. **Find LinkedIn profile** via web search
2. **Scrape LinkedIn posts** (if available via Apify)
3. **Scrape X (Twitter) posts** — if user provided an X username, run:

```bash
apify call apidojo/tweet-scraper --input '{
  "searchTerms": ["from:USERNAME"],
  "maxTweets": 10,
  "sort": "Latest"
}' --output-dataset --silent
```

Replace `USERNAME` with the target's X handle (without `@`).

- Parse the dataset output and extract: tweet text, date, likes, retweets, replies
- If `APIFY_API_TOKEN` is not set or scraping fails, fall back to web search for recent X posts
- Skip this step silently if no X username was provided

4. **Gather background info**:
   - Previous work experience
   - Education
   - Notable projects/achievements
   - Recent posts/topics (from both LinkedIn and X)

### Step 4: Generate Comprehensive Playbook

Follow the **Full Playbook Template** below with ALL sections:
- Target Profile Summary
- Company & Industry Research (detailed)
- Founder Deep Dive (for startup targets)
- Recent Content & Posts
- Profile Comparison
- 8 Questions
- 4 Talking Points
- Communication Tips
- Pre-Chat Checklist

### Step 5: Save & Push to Notion (optional)

1. **Save locally**: `memory/coffee-chat-{target-slug}-{YYYY-MM-DD}.md`
2. **Push to Notion**: Append to your configured Notion page (if `NOTION_API_KEY` is set)

---

## Full Playbook Template

```markdown
# ☕ Coffee Chat Playbook: [Target Company/Name]

**Date:** [YYYY-MM-DD]
**Status:** Researched

---

## 📋 Target Profile Summary

### Basic Info
- **Target**: [Name] (Founder/CEO/PM at [Company])
- **LinkedIn**: [URL]
- **Location**: [City, Country]
- **Connection**: [How you met]

### Background
- [Previous career/experience]
- [Notable projects or achievements]
- [Education]

### Recent Posts / Topics
- [Topic 1 from LinkedIn/X]
- [Topic 2]
- [Topic 3]

---

## 🏢 Company & Industry Research

### Company Overview
- **Company Name**: [Name]
- **Founded**: [Year]
- **Location**: [HQ location]
- **Funding**: [Amount] ([Investors]) - [Date]
- **Team**: [Size/distribution if known]
- **X/Website**: [Links]

### Product
- [Product description]
- [Key features]
- [Target users]

### Industry Landscape
- **Market**: [Brief description of market]
- **Competitors**: [List key competitors]
- **Trends**: [Industry trends]
- **Challenges**: [Key challenges]
- **Opportunities**: [Opportunities]

### Investors (if applicable)
- **Investor 1**: [Background]
- **Investor 2**: [Background]

---

## 👥 Founder Deep Dive (for startups)

### [Founder 1 Name] ([Title])
- **Background**: [e.g., Documentary filmmaker, Ex-Google, etc.]
- **Work history**: [Notable companies/clients]
- **Notable achievements**: [e.g., Founded X, Award Y]
- **Philosophy/Interests**: [From posts]

### [Founder 2 Name] ([Title])
- [Same structure]

---

## 📱 Recent Content & Posts

### LinkedIn Posts
- [Post 1: topic and key message]
- [Post 2]
- [Post 3]

### X (Twitter) Posts — Last 10
> Scraped via apidojo/tweet-scraper. If X username not provided, use web search results.

| # | Date | Tweet | Likes | RT | Replies |
|---|------|-------|-------|----|---------|
| 1 | [YYYY-MM-DD] | [Full tweet text] | [n] | [n] | [n] |
| 2 | [YYYY-MM-DD] | [Full tweet text] | [n] | [n] | [n] |
| 3 | [YYYY-MM-DD] | [Full tweet text] | [n] | [n] | [n] |
| ... | | | | | |

**Key themes from X posts:**
- [Theme 1: what they talk about most]
- [Theme 2: opinions/stances]
- [Them