Agent-Native Operator

The skills, subagents, and identity files for the Agent-Native Operator Bootcamp by LevelUp Labs.

This repo bundles everything you need to follow along with the workshop. Pre-work walks you through standing up OpenClaw on a VPS; the workshop installs everything in this repo on top.

---

What's inside

agent-native-operator/
├── prework/                 Read this before the workshop
│   ├── README.md            Setup walkthrough (~45-60 min)
│   └── security_check.md    Detailed security checks the bot reads
│
├── identity/                Installed during the workshop
│   ├── SOUL.md              Chief of Staff persona (name-agnostic)
│   ├── USER.md              Template — fill in during the workshop
│   └── AGENTS.md            Operating manual
│
├── skills/                  Installed during the workshop
│   ├── daily_brief/         The capstone automation
│   ├── skill_review/        Weekly heartbeat that improves your skills
│   └── build-it/            Meta-skill — scaffolds new automations
│
├── agents/                  Named subagents
│   └── research_worker/     Parallel research worker
│
├── slides/                  Lecture content (for reference)
│   ├── lecture_1.md
│   └── lecture_2.md
│
├── worksheets/              Co-working session worksheets
│   ├── coworking_1.md
│   └── coworking_2.md
│
├── install.sh               Run this during the workshop
└── README.md                You are here

---

Pre-workshop (do this BEFORE Saturday)

Read prework/README.md. It walks you through:

1. Getting an LLM API key
2. Deploying OpenClaw on a Hostinger VPS
3. Running the security check
4. Naming your Claw

Estimated time: 45-60 minutes. If you hit issues, the troubleshooting section in the pre-work guide covers the common ones.

---

During the workshop

The workshop content (identity files, skills, sub-agents, demo papers) needs to land in your OpenClaw workspace. Two paths produce the same result:

Primary path — conversational install. Once OpenClaw is running, paste this into your web chat:

Pull the workshop content from github.com/levelup-labs-ai/agent-native-operator into my workspace. Copy the identity files (SOUL.md, USER.md, AGENTS.md), all the skills under skills/, the sub-agents under agents/, and the demo papers under wiki/raw/. Seed an empty feedback.md alongside each skill. Confirm what was installed when you're done.

Your Claw pulls the repo, places the files correctly, and reports back. This is the recommended path because it teaches the conversational mode you'll use for the rest of the workshop.

Fallback path — install.sh. If you'd rather run a script, clone the repo onto your VPS and run install.sh:

git clone https://github.com/levelup-labs-ai/agent-native-operator.git
cd agent-native-operator
./install.sh

After install (either path):

• The Chief of Staff persona is loaded (you'll customize your Claw's name in pre-work; the persona itself is name-agnostic)
• Skills are available: daily_brief, inbox_watch, meeting_summary, plan_my_day, skill_review, build-it, wiki_compile. Each has an empty feedback.md seeded.
• Sub-agent paper_reader is registered (used in the Lecture 2 sub-agent demo).
• The 25 prompt-engineering demo papers are in wiki/raw/. The pre-compiled wiki ships in wiki/.

You'll then customize USER.md live in Lecture 1 and fill in the "My setup" section of each skill's SKILL.md conversationally — your Claw asks the personalization questions and proposes SKILL.md edits for your approval. MEMORY.md stays empty in Lecture 1 (it grows organically over the following weeks as your Claw learns patterns from your daily logs and corrections).

---

What you're building toward

By the end of the workshop, you'll have:

• A proactive Chief of Staff agent running on your own infrastructure
• One working scheduled automation (your daily brief, on your sources)
• A weekly improvement loop (skill_review) that gets sharper over time as you give it feedback
• The /build-it meta-skill for scaffolding any future automation

You leave with the framework, the artifacts, and a 30-day ramp plan.

---

After the workshop

Use /build-it whenever you have an idea for a new automation. The meta-skill walks you through the same 5-step framework you learned in the workshop, every time.

---

Course

Agent-Native Operator Bootcamp — LevelUp Labs https://maven.com/aishwarya-kiriti/openclaw-bootcamp