Marketing / playbook

Build a Threads system that learns from every post.

Use your own performance data to find ideas, draft posts, publish deliberately, and improve the next batch.

KKeith Rumjahn·9 min read·Free

THE FEEDBACK LOOP · ILLUSTRATIVE

Each cycle gives the next draft better evidence.

One reviewed workflow—not an autopilot promise

Scout: Collect useful source material from the inputs you approved.
Draft: Compare ideas with your own performance patterns, then create candidates for review.
Publish: Approve the exact post and destination before scheduling through the configured integration.
Learn: Sync real results, check the sample and date range, and update what the next cycle knows.

An illustrative operating loop, not a claim about follower or view growth. The repository and connected platforms determine which steps are available.

This is not a prompt that writes random posts. It is a feedback loop: collect useful source material, compare it with what has worked for you, draft in your voice, publish deliberately, and feed real performance back into the next round.

YOU’LL WALK AWAY WITH

The public repository, a safe setup plan, and a copyable prompt for adapting the system to your accounts.

GitHub · Codex or Claude Code · Python · Postiz for scheduling

01 / Build a loop, not a post generator

The useful part is not asking AI to “write a viral Thread.” The system keeps a record of your published posts and their results, uses those results to identify patterns worth testing, and gives the next draft better context.

The current project is called Orbit. It can sync performance data from Threads, LinkedIn, and YouTube; discover source material; create drafts; schedule through Postiz; and store the results locally so the next cycle can learn from evidence.

Scout: collect promising ideas from sources you choose.
Draft: compare ideas with your own stronger posts and write a new version.
Publish: review every draft and schedule only what you approve.
Learn: sync the results and update the evidence for the next batch.

02 / Start with the public repository

The repository is public, so you can inspect or download it without creating an account here. Its current Orbit architecture is the source of truth. Some top-level instructions describe the older Hermes skill workflow; they are useful history, but they do not replace the live architecture and code.

Orbit is a technical, local-first Python project rather than a one-click web app. The full workflow currently uses Postiz for scheduling. Adapting another scheduler requires implementation work; do not assume it is already connected.

BUILT BY KEITH

Threads Growth Skill on GitHub

Open the source, architecture, setup notes, tests, and current Orbit CLI.

Open the public repository

03 / Keep the first run safe

Use Codex or Claude Code inside a dedicated checkout. Let it inspect the repository instructions and current commands before proposing setup. Never paste API keys, cookies, or account credentials into a chat; put secrets only in the environment or configuration location the current code expects.

Start read-only: configure the platforms you choose, sync data, inspect status and insights, and generate drafts without publishing. Review the output yourself. Require explicit approval before anything is scheduled or posted.

Preserve any existing checkout and uncommitted work.
Choose only the platforms and data sources you actually want to connect.
Run the repository’s complete test suite before trusting a change.
Do not automate replies, follows, engagement, or publishing without a separate review.

READY TO MAKE IT YOURS?

Get the exact prompt.
Make it your own.

You’ve seen how it works. Unlock the copyable prompt and the remaining steps to try it yourself.

Copy the setup prompt
Prove one cycle before scaling it

New here? Create a free account

Already a member? Use your existing account. Every playbook is free.