Penny — Build-in-Public Content Agent

Most founders know they should be building in public. Almost none do it consistently — it takes time they don't have.

Penny runs on a schedule. You tell her what you shipped, learned, or decided. She turns it into three pieces of content — a full blog post, a LinkedIn post, and a Twitter thread — all in your brand voice. Publish or queue. Move on.

Varconi uses Penny internally. Every agent shipped, every lesson learned — Penny captures it.

---

What Penny produces per run: → Blog post (800–1,200 words): structured narrative, your voice, SEO-aware slug and meta → LinkedIn post: professional framing, hook + insight + CTA, under 1,300 characters → Twitter/X thread: 6–10 tweets, punchy, shareable, designed for engagement → All three saved as markdown files to content/ready/ — ready to publish

---

What makes Penny different from a prompt:

• Remembers your brand voice across runs via SOUL.md
• Pulls from your real build context — specific, not generic
• Knows the format difference between LinkedIn, Twitter, and long-form
• Output goes to structured files, not a chat window you copy from
• Runs autonomously on a schedule — content never falls behind your build

---

Output format:

• content/ready/YYYY-MM-DD-blog.md
• content/ready/YYYY-MM-DD-linkedin.txt
• content/ready/YYYY-MM-DD-twitter-thread.txt

---

Requires: OpenClaw, DeepSeek API key (~/bin/zsh.001/run) Compatible with: Ghost, Substack, Buffer, or any publishing workflow.