Every Prompt You Send Drags 18,384 Words Of Junk. Here’s How I Cut It.

Every Prompt You Send Drags 18,384 Words Of Junk. Here’s How I Cut It.

More

Descriptions:

Full post w/ Clean Your Harness Guide: https://natesnewsletter.substack.com/p/ai-harness-audit?r=1z4sm5&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Your new AI model is running on instructions you wrote for the old one. This is the AI harness audit: how to find the hidden setup shaping your AI and clean it safely.

I mapped every skill, memory, system prompt, permission, and check around my own AI and found 66 skill routes and 172 instruction files quietly slowing the models down. In this deep dive I test a compact setup against a much heavier one, show why the richer prompt still lost on delivery, and give you six rules for a harness that survives the next model upgrade. Then I hand you the cleaner so you can run the same audit on your own Claude or Codex setup.

My Links 🔗
👉🏻 Newsletter: https://natesnewsletter.substack.com/
👉🏻 X: https://x.com/natebjones
👉🏻 TikTok: https://www.tiktok.com/@nate.b.jones
👉🏻 Instagram: https://www.instagram.com/nate.b.jones

What’s really happening inside your AI setup?
The common story is that a new model just makes everything better, but the real question is what that model inherits from the setup you already built.

In this video, I share the inside scoop on auditing the harness around your AI:

– Why your upgraded model can feel worse than a fresh chat
– How a compact brief beat a richer method on delivery
– What six rules keep a harness clean as models change
– Where Fable 5 and ChatGPT 5.6 fail for different reasons

A cleaner harness won’t make a weak model strong, but it will tell you whether the problem is the model or the machinery you wrapped around it.

Chapters:
00:00 I Overbuilt My AI Harness
01:58 The Hidden Cost of Harness Bloat
04:54 Rule 2: Blame the Right Layer
06:06 Rule 3: One Rule, One Home, One Owner
07:27 Rule 4: Load It When You Need It
08:25 Rule 5: Hard Rules Need Hard Checks
08:58 Rule 6: Build for the Model and Product
10:58 The Codex Audit Results
12:17 Fable 5 vs. ChatGPT 5.6 Failure Modes
12:54 A Clean Harness Without the Barnacles
14:17 Own the System Around Your AI
14:39 The Cleaner Skill

Listen to this video as a podcast.
– Spotify: https://open.spotify.com/show/0gkFdjd1wptEKJKLu9LbZ4
– Apple Podcasts: https://podcasts.apple.com/us/podcast/ai-news-strategy-daily-with-nate-b-jones/id1877109372

1 Item

Channels

1 Item

People