Who it is forA knowledge worker or builder who needs a repeatable domain agent but does not want to write its instructions by hand.
The problemWriting a complete operating manual with edge cases and decision rules takes long, careful effort.
The outcomeA saved agent with a drafted manual, a record of which model drafted it, and a Neural tab showing ongoing learning.
What you need first
- Docker with Docker Compose, or a free Google account for the Colab route
- At least one provider API key, such as OpenRouter, Google AI Studio or NVIDIA NIM
- A browser to open the app and paste keys into the Vault
Steps
- Start ENZO with
docker compose up -dand open http://localhost:5001 after pasting a provider key. - Open the Agent builder and describe the task in plain English, for example an agent that researches a specific domain.
- Wait for the two-pass drafter to finish and review the operating manual it writes.
- Open the agent's Neural tab to watch NEURAL FOCUS and the learning layer update over time.
Prompt for your agent
Using the repo theguysudo/ENZO running at http://localhost:5001, create an agent from this task: <task-description>. Expected output: a saved agent with a full operating manual and a recorded drafting model. Verify the agent page names the model that actually drafted it.
Caveat. The README says the agent builder races free candidates from your own providers, so drafts depend on which free-tier models are reachable at that moment.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for theguysudo/ENZO and "Draft an agent's operating manual from one sentence"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer