Lead Magnet

Lead Magnet Skill

Create high-converting, story-driven lead magnets in your brand voice, write them as Notion-ready content, optionally publish to Notion, and draft LinkedIn promo posts.

Personalization

This skill works in your brand voice. Before running, load the brand profile at ~/.claude/brand-profile.md.

• If it doesn't exist, run the personalize skill first (or just tell me your name, niche, offer, and audience — I'll create it).
• If a field this skill needs is blank, I'll ask one or two quick questions and save the answers back to the profile so you only answer once.
• Never invent facts or results about the user — use only what's in the profile, or ask.

The steps below read these brand-profile fields: Name, Audience, Main-pain, Offer, Transformation, Tone, Words-to-avoid, and Default-CTA. Wherever a step needs a name, audience, offer, proof point, or CTA link, pull it from the profile — never hardcode.

Requirements

• Content generation (core value): No credentials needed. Researching, writing the JSON, and drafting the LinkedIn posts all work with no setup.
• Notion push (optional): The scripts/push_to_notion.py step needs two environment variables:
• NOTION_TOKEN — your Notion internal integration secret
• NOTION_DATABASE_ID — the ID of the Notion database to add pages to
• If either is not set, skip the push and tell the user exactly what to add (see Step 5). The lead magnet content is still fully delivered as a JSON file they can push later.

Workflow

1. Parse user input for topic, format, and any special instructions
2. Fetch YouTube transcript (if a YouTube URL was provided)
3. Research the topic (mandatory — see references/voice-and-rules.md "Research First")
4. Write the lead magnet as a Notion-ready JSON file
5. Push to Notion (optional) using scripts/push_to_notion.py
6. Draft 3 LinkedIn post variations with different hooks
7. Present everything to the user for review

Step 1: Parse Input

Extract from the user's message:

• TOPIC: What the lead magnet is about (if a YouTube URL is given, derive the topic after fetching the transcript)
• FORMAT: One of: promptpack, playbook, systemprompts, framework, swipe_file, guide (default: guide)
• SPECIAL INSTRUCTIONS: Any constraints, angles, or preferences mentioned

Display your parsing:

Creating lead magnet:
- Topic: {TOPIC}
- Format: {FORMAT}
- Notes: {SPECIAL INSTRUCTIONS or "none"}

Step 2: Fetch YouTube Transcript (if applicable)

If the user provided a YouTube URL, extract the video ID and fetch the transcript:

python3 -c "
from youtube_transcript_api import YouTubeTranscriptApi
api = YouTubeTranscriptApi()
transcript = api.fetch('VIDEO_ID')
text = ' '.join([t.text for t in transcript])
print(text)
"

Use the transcript as the primary source for the lead magnet topic and content. Extract:

• The core topic/thesis of the video
• Key frameworks, steps, or insights mentioned
• Specific numbers, tools, or examples used
• The angle to take on the subject (in your Tone from the brand profile)

Then update your parsing display with the derived topic.

Step 3: Research (MANDATORY)

Before writing ANY lead magnet, deeply research the topic. Use WebSearch and WebFetch to understand:

• What real people are saying on Reddit, X, YouTube
• How people actually use/implement the topic
• Common pain points and questions (anchor these to the Main-pain field in the brand profile)
• Specific tools, numbers, and examples

Never write from assumptions. See references/voice-and-rules.md for full research requirements.

Step 4: Write the Lead Magnet JSON

Create a JSON file in ~/lead-magnets/output/ (create the folder if it doesn't exist), following the exact structure documented in references/json-format.md. Use a kebab-case filename derived from the topic.

Critical rules (see references/voice-and-rules.md for complete list):

• 1,500 words MAX — hard constraint
• 6th grade reading level — radically simple
• ZERO markdown bold (**text**) in JSON content strings — Notion API shows raw asterisks
• Story-driven — every lead magnet needs a narrative arc (BAB, Origin Story, Case Study Chain, or Behind the Curtain)
• Concrete proof — real numbers, specific examples, named frameworks (use only proof points from the brand profile, or ask — never invent)
• Early CTA — purple callout with 🚀 after intro, linking to the Default-CTA from the brand profile
• Closing CTA — at bottom, also linking to the Default-CTA
• Never invent revenue, follower counts, or results for the user

Step 5: Push to Notion (Optional)

First check that the required environment variables are set:

if [ -z "$NOTION_TOKEN" ] || [ -z "$NOTION_DATABASE_ID" ]; then
  echo "SKIP_NOTION"
fi

• If it prints SKIP_NOTION, skip this step. Tell the user:

> "I've saved the lead magnet JSON. To publish it to Notion, set NOTION_TOKEN (your Notion internal integration secret) and NOTION_DATABASE_ID (the target database ID) in your environment, then run: python3 scripts/push_to_notion.py "

• Otherwise, run the push script:

python3 "${SKILL_ROOT}/scripts/push_to_notion.py" ""

The script reads NOTION_TOKEN and NOTION_DATABASE_ID from the environment. If the push fails, show the error and troubleshoot.

Step 6: Draft 3 LinkedIn Post Variations

Write 3 LinkedIn posts promoting the lead magnet, in the Tone from the brand profile and respecting Words-to-avoid. Each must have a different hook (first 2 lines).

LinkedIn post rules:

• 150-200 words each
• First 2 lines are the most important — they must stop the scroll
• Use white space between sentences for mobile readability
• End with a comment-trigger CTA (e.g., "Comment KEYWORD and I'll send it")
• Include 3-5 hashtags
• Tone: pull from the brand profile
• Each variation should use a different angle:
• Variation A: Contrarian or hot-take hook
• Variation B: Pain-first / problem-aware hook (anchor to Main-pain)
• Variation C: Results-led / proof hook (using only real proof points)

Step 7: Present to User

Show:

1. Confirmation of the JSON file path (and the Notion page ID if it was pushed)
2. All 3 LinkedIn post variations clearly labeled
3. Ask if they want any tweaks before posting

Content Categories

1. Prompt Packs — Copy-paste prompts organized by goal, each with WHY it works + example output
2. AI Playbooks — Step-by-step guides, story-driven, proof-backed
3. System Prompt Templates — Ready-to-paste AI configurations
4. Frameworks & Blueprints — Visual systems showing HOW something works
5. Swipe Files — Curated collections with commentary on WHY each works
6. Guides — Deep-dive on one topic, story-first

What We Do NOT Create

• Automation tutorials (no n8n, Make.com, Zapier how-tos)
• Tool-dependent content (must work regardless of AI tool)
• Generic guides with no unique data
• Lists without narrative
• Content requiring technical skill to use

Reference Files

• references/voice-and-rules.md — Complete voice rules, storytelling frameworks, proof standards, design system, quality checklist
• references/json-format.md — Exact JSON structure with block type examples
• references/linkedin-examples.md — Example LinkedIn post structure for reference

Read these references BEFORE writing. They contain critical formatting rules that prevent production issues.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: duncan-buildroom
• Source: duncan-buildroom/profit-room-skills
• License: MIT

Install and usage instructions live in the source repository linked above.