dropspace-content-engine

name	dropspace-content-engine
description	Self-improving content pipeline for Dropspace. Analyzes post performance, generates new content across 6 platforms, schedules publishing, and compounds strategy over time. Run nightly for best results. Use when asked to run the content engine, generate social posts, or manage the content pipeline.
homepage	https://www.dropspace.dev/community/dropspace-content-engine
source	https://github.com/joshchoi4881/markus
requires	{"env":["DROPSPACE_API_KEY","ANTHROPIC_API_KEY","FAL_KEY"],"install":"git clone https://github.com/joshchoi4881/markus && cd markus && npm install"}

AI Content Engine

Self-improving content pipeline. Run two scripts nightly — posts get smarter over time because the pipeline learns from real engagement data.

Setup

1. Clone and install

git clone https://github.com/joshchoi4881/markus.git
cd markus && npm install

If canvas fails to install, that's fine — text-only formats still work. For visual formats (TikTok/Instagram slideshows), install the system dependencies first: macOS brew install pkg-config cairo pango, Linux apt install libcairo2-dev libpango1.0-dev.

2. Set your API keys

export DROPSPACE_API_KEY="ds_live_..."   # from dropspace.dev/settings/api
export ANTHROPIC_API_KEY="sk-ant-..."           # from console.anthropic.com
export FAL_KEY="fal_..."                        # from fal.ai (for visual/video formats)

Save in a .env file (copy from templates/.env.example). Add .env to .gitignore to avoid committing secrets.

3. Initialize your app

Replace myapp with your app name. Add or remove platforms as needed.

node scripts/init-app.js --app myapp --platforms tiktok,twitter,linkedin,instagram,reddit,facebook

This creates apps/myapp/ with config files and per-platform directories.

4. Edit app.json

Open apps/myapp/app.json and fill in:

• name — your product name
• description — one-sentence description
• audience — who you're targeting (e.g., "indie hackers, solo founders")
• problem — the pain your product solves
• differentiator — why you vs alternatives
• url — your product URL
• voice — writing style (e.g., "lowercase, casual, genuine")

5. Validate

node scripts/test-pipeline.js --app myapp

This checks: env vars present, API keys valid, app.json readable, and runs a dry self-improve cycle. All checks should pass before running the real pipeline.

Run the Pipeline

The core loop — run these two commands:

# Analyze performance + generate new posts (up to 60 min)
node scripts/run-self-improve-all.js --app myapp

# Schedule generated posts for today (up to 40 min)
node scripts/schedule-day.js --app myapp

What Happens Each Run

1. Pulls 14 days of analytics from Dropspace API
2. Identifies winning hooks (posts with >2x average engagement)
3. Kills underperforming format experiments
4. Generates 7-14 new posts with AI (text, visual, video formats)
5. Writes strategy notes to apps/myapp/ (persists between runs)
6. Schedules posts across the day via Dropspace API

Each run reads the previous run's strategy notes. The loop compounds.

Automate It

Set up system crons to run the full nightly cycle (replace paths and app name):

crontab -e

# Full nightly schedule (all times ET):
0  0 * * * cd /path/to/markus && source .env && node scripts/refresh-tracking.js --app myapp && node scripts/cleanup-posts.js --app myapp --days 7 >> /tmp/dropspace-cron.log 2>&1
30 0 * * * cd /path/to/markus && source .env && node scripts/run-x-research.js --app myapp >> /tmp/dropspace-cron.log 2>&1
0  1 * * * cd /path/to/markus && source .env && node scripts/run-self-improve-all.js --app myapp >> /tmp/dropspace-cron.log 2>&1
0  2 * * * cd /path/to/markus && source .env && node scripts/schedule-day.js --app myapp >> /tmp/dropspace-cron.log 2>&1

Time	Script	What it does
12:00 AM	refresh-tracking.js + cleanup-posts.js	refresh analytics cache, clean old media
12:30 AM	run-x-research.js	scan X/Twitter for trending hooks (optional, needs Bird CLI)
1:00 AM	run-self-improve-all.js	analyze performance, generate new posts (up to 60 min)
2:00 AM	schedule-day.js	schedule generated posts for today

The x-research cron is optional — skip it if you don't have Bird CLI set up. The other 3 are the core pipeline.

Or run manually each day — the pipeline still compounds as long as you run it regularly.

Content Formats

Format	Type	Platforms
story-slideshow	visual	TikTok, Instagram, Facebook
ugc-reaction	video	TikTok, Instagram
ugc-talking	video	TikTok, Instagram
text-single	text	Twitter/X
text-post	text	LinkedIn, Reddit, Facebook

Error Recovery

• Add --dry-run to any script to preview without saving
• schedule-day.js is idempotent — safe to re-run same day
• Failed platforms retry automatically on next self-improve run
• Check logs at apps/myapp/reports/