competitor-content-intel

name	competitor-content-intel
description	Weekly competitor tracking + content pipeline. Scans YouTube and X for creators in your niche, finds the top trending topics, writes full video scripts in your voice, drafts one X article, creates tasks in Todoist, and posts a summary to Discord. One run, full job.

Competitor Content Intel → Full Content Pipeline

One run does everything: scrape competitors → identify trending topics → write video scripts in your voice → draft X article → create Todoist tasks → post Discord summary.

Setup

Before running, configure the following in your environment:

# Required API keys (set in your .secrets/ directory or as env vars)
X_BEARER_TOKEN=           # Twitter/X API v2 bearer token (developer.twitter.com)
NOTION_API_KEY=           # Notion integration token (notion.so/my-integrations)
TODOIST_API_KEY=          # Todoist REST API token (todoist.com/app/settings/integrations)
DISCORD_CHANNEL_ID=       # Discord channel ID to post summaries to
NOTION_SCRIPTS_PARENT_ID= # Notion page ID where video scripts will be created
NOTION_SWIPE_FILE_ID=     # Notion page ID containing your voice/style swipe files (optional)
TODOIST_PROJECT_ID=       # Todoist project ID for content tasks
TODOIST_SECTION_ID=       # Todoist section ID (optional)

Competitors to Track

Edit this list to match your niche. The default list is for the AI tools / coding / indie hacking space:

Creator	YouTube Channel	X Handle
Alex Finn	@alexfinndev	@alexfinndev
NoCode MBA	@nocodemba	@nocodemba
Drake Surach	Search "Drake Surach AI"	@drakesurach
Matt Berman	@MattVidPro	@mattberman_
Cole Medin	@ColeMedin	@ColeMedin
Nick Dobos	@NickDobos	@NickDobos
McKay Wrigley	@McKayWrigley	@mckaywrigley
Greg Isenberg	@GregIsenberg	@gregisenberg
Pieter Levels	@levelsio	@levelsio
Andrej Karpathy	@AndrejKarpathy	@karpathy
Riley Brown	@rileybrown_ai	@rileybrown_ai
Nick St. Pierre	Search "Nick St Pierre AI"	@nickstPierre
Ammaar Reshi	@ammaar	@ammaar
David Shapiro	@DavidShapiroAI	@DavidShapiroAI
Sam Witteveen	@samwitteveenai	@samwitteveen
The AI Advantage	@TheAIAdvantage	@AI_Advantage
Fireship	@Fireship	@fireship_dev
NetworkChuck	@NetworkChuck	@NetworkChuck
Tina Huang	@TinaHuang1	@_tinahuang
All About AI	@AllAboutAI	@allabtai
Unconventional Coding	@UnconventionalCoding	—
Logan Kilpatrick	—	@OfficialLoganK

Add your own niche-specific creators to this table. The more targeted your list, the better the topic recommendations.

Your Channel Profile

Update this section to reflect your channel before running:

Channel: YOUR_YOUTUBE_HANDLE
Subscribers: X
Niche: [e.g. AI tools, SaaS, coding tutorials, crypto]
Top performing content: [e.g. tool comparisons, hands-on builds, tutorials]
Style: [e.g. casual, technical, documentary]
X handle: YOUR_X_HANDLE
X followers: X

The agent uses this profile when writing scripts to make sure the angle matches YOUR voice and competitive position — not a generic creator.

---

Step 1 — YouTube Research (last 7 days only)

For each competitor, use web_fetch on their YouTube channel's videos tab: https://www.youtube.com/@{handle}/videos

Extract: video title, view count, publish date, URL. Only include videos published in the last 7 days. Skip anything older.

Pick the 1-2 highest-view videos per creator. If a creator posted nothing in the last 7 days, skip them.

Step 2 — X Research (last 7 days, X API only)

Use the X API directly. Do NOT use web_search for X data — it returns unreliable results.

# Broad topic sweep — customize these keywords for your niche
curl -s -X GET "https://api.twitter.com/2/tweets/search/recent?query=YOUR_KEYWORD_1%20OR%20YOUR_KEYWORD_2%20OR%20YOUR_KEYWORD_3&max_results=50&tweet.fields=public_metrics,created_at,author_id&sort_order=relevancy" \
  -H "Authorization: Bearer $X_BEARER_TOKEN"

# Per-creator lookup
curl -s -X GET "https://api.twitter.com/2/tweets/search/recent?query=from%3A{handle}&max_results=10&tweet.fields=public_metrics,created_at&sort_order=relevancy" \
  -H "Authorization: Bearer $X_BEARER_TOKEN"

Pull top posts by likes. Note topic, engagement, and why it performed.

Step 3 — Identify Top 3 Topics

After pulling YouTube + X data, identify the 3 topics with the most engagement this week. For each:

• What is the specific angle?
• Why is it performing? (novelty, controversy, tutorial demand)
• What's YOUR unique version? (what do you have that they don't — access, receipts, live builds, real numbers)

Write to a local state file for handoff to downstream steps:

mkdir -p ./state
python3 << PYEOF
import json, datetime
topics = [
    {"title": "TOPIC_1", "source": "SOURCE / PLATFORM", "url": "URL"},
    {"title": "TOPIC_2", "source": "SOURCE / PLATFORM", "url": "URL"},
    {"title": "TOPIC_3", "source": "SOURCE / PLATFORM", "url": "URL"},
]
data = {"updatedAt": datetime.datetime.utcnow().strftime('%Y-%m-%dT%H:%M:%SZ'), "topics": topics}
json.dump(data, open('./state/trending-topics.json', 'w'), indent=2)
print("Written")
PYEOF

Step 4 — Load Your Voice (MANDATORY before writing anything)

Before writing a single word:

1. Read your voice/style guide (a SKILL.md or markdown file describing your tone, vocabulary, sentence structure, hooks, and what you never do)
2. Pull your swipe files — examples of your best performing content

# If stored in Notion:
curl -s "https://api.notion.com/v1/blocks/$NOTION_SWIPE_FILE_ID/children" \
  -H "Authorization: Bearer $NOTION_API_KEY" \
  -H "Notion-Version: 2022-06-28"

Apply hook patterns and structural patterns from the swipe files. If you skip this step, the scripts will sound generic — the whole point of this pipeline is content that sounds like you, not like an AI wrote it.

Step 5 — Write Video Scripts in Notion (3 scripts)

For each of the 3 topics, create a Notion page under $NOTION_SCRIPTS_PARENT_ID.

Script page structure:

• Date + Why Now — today's date, why this topic is hot this specific week
• Title — punchy, search-optimized YouTube title
• Hook (first 30 seconds) — must match a pattern from your swipe files. Never generic.
• Outline — 6-8 bullet points (talking points, not word-for-word)
• CTA — what to say at the end
• Thumbnail concept — one sentence
• Why it'll perform — your unique angle vs the competitor version
• Original inspiration — embed the competitor video/tweet that sparked the idea

Embedding Links (MANDATORY for tweets and YouTube links)

Never paste raw URLs. Use Notion embed blocks:

{
  "object": "block",
  "type": "embed",
  "embed": { "url": "TWEET_OR_YOUTUBE_URL" }
}

Use "type": "embed" for Twitter/X posts and YouTube videos. Use plain paragraph text for regular article links.

Notion API — Create Script Page:

curl -s -X POST "https://api.notion.com/v1/pages" \
  -H "Authorization: Bearer $NOTION_API_KEY" \
  -H "Notion-Version: 2022-06-28" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"page_id": "'"$NOTION_SCRIPTS_PARENT_ID"'"},
    "properties": {"title": {"title": [{"text": {"content": "TITLE"}}]}},
    "children": [BLOCKS]
  }'

Save the returned page ID for each script.

Step 6 — Write X Article Draft (ONE only — best topic)

Pick the ONE topic with the most depth, strongest opinion, or clearest how-to that reads well as a standalone post. Ask: "could someone learn something real just from reading this?" If yes, that's the one.

X Format Rules (no exceptions):

• NEVER write a thread — one post only, always
• NEVER use hashtags
• NEVER add links in the body
• Write in your authentic voice — match your swipe files exactly
• No corporate language, no hedging, no AI slop phrases

X Article Format:

Long-form post — no character limit on X articles. Write as long as it takes to do the topic justice.

• Hook first line — make someone stop scrolling
• Multiple short paragraphs, each its own line break
• Technical but readable — explain the thing properly
• Real numbers, real examples, real context
• Punchy one-line closer

Quality Rules:

• No artificial length caps — stop when you've said everything worth saying
• Go deep technically — explain how it actually works
• Never "here's how" without actually explaining how
• Minimum 800 characters — if shorter, you haven't said enough
• Teach something concrete — reader should know something they didn't before

Paid Tier Suggestion (optional)

If you have a paid subscription tier, add a 💰 callout at the bottom of each script page suggesting what to put behind the paywall for that topic.

The rule: 99% is free (the concept, the explanation, the how-to). Paid = the actual artifact (the file, template, config, exact prompt, the thing they'd spend hours building themselves).

Create X article as child page under the chosen script page:

curl -s -X POST "https://api.notion.com/v1/pages" \
  -H "Authorization: Bearer $NOTION_API_KEY" \
  -H "Notion-Version: 2022-06-28" \
  -H "Content-Type: application/json" \
  -d '{
    "parent": {"page_id": "CHOSEN_SCRIPT_PAGE_ID"},
    "properties": {"title": {"title": [{"text": {"content": "[X ARTICLE] TITLE"}}]}},
    "children": [BLOCKS]
  }'

Step 7 — Create Todoist Tasks

# 3 script tasks (one per video):
curl -s -X POST "https://api.todoist.com/rest/v2/tasks" \
  -H "Authorization: Bearer $TODOIST_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "🎬 TITLE",
    "description": "Script: NOTION_URL",
    "project_id": "'"$TODOIST_PROJECT_ID"'",
    "section_id": "'"$TODOIST_SECTION_ID"'",
    "due_string": "today"
  }'

# 1 X article review task:
curl -s -X POST "https://api.todoist.com/rest/v2/tasks" \
  -H "Authorization: Bearer $TODOIST_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "content": "🐦 review x article — CHOSEN_TITLE",
    "description": "X article draft: X_DRAFT_URL",
    "project_id": "'"$TODOIST_PROJECT_ID"'",
    "due_string": "today"
  }'

Step 8 — Post to Discord

Post a single message to $DISCORD_CHANNEL_ID.

Format — links only, no content dumped into Discord:

content pipeline ran ✅

**{topic 1 title}** — 📝 {notion_url}
**{topic 2 title}** — 📝 {notion_url}
**{topic 3 title}** — 📝 {notion_url} + 🐦 x article: {x_draft_url}

nothing posted. x article ready for review.

---

Rules

• 7 days max — nothing older. Trending content is perishable.
• X API only — never web_search for X/Twitter data.
• Load your voice before writing — every single run, no exceptions.
• Embed tweets and YouTube links — never raw URLs in Notion.
• NEVER write a thread — one X post only, always.
• NEVER post to X directly — Notion drafts only. Human reviews and approves.
• Max 3 topics per run — quality over quantity.
• One X article draft — pick the best topic, write one post.
• Save script IDs to state file before posting to Discord — ensures output is recoverable if the session dies.
• ONE message to Discord — combine all output into a single message.
• 10 minute hard limit — if not done in 10 minutes, post whatever you have and stop.

Running as a Cron

This skill is designed to run daily via a cron job. Example OpenClaw cron config:

{
  "name": "competitor-content-intel",
  "schedule": { "kind": "cron", "expr": "0 10 * * *", "tz": "America/Chicago" },
  "payload": {
    "kind": "agentTurn",
    "message": "Run the competitor-content-intel skill. Research competitors, identify top 3 topics from the last 7 days, write 3 video scripts and 1 X article draft to Notion, create Todoist tasks, post summary to Discord.",
    "timeoutSeconds": 600
  },
  "sessionTarget": "isolated"
}

File Structure

competitor-content-intel/
├── SKILL.md          ← this file
└── state/
    ├── trending-topics.json   ← output from Step 3, input to content-factory
    └── script-ids.json        ← Notion page IDs written after Step 5

The state/trending-topics.json file is designed to be consumed by a downstream content-factory skill or cron if you want to chain pipelines.