# Hermes Agent Mastery: An AI Shift That Runs the Work for You

URL: https://ongchummo.vn/en/courses/hermes-agent-mastery

Master Hermes Agent in 6 modules: write skills reusable across harnesses, run via chat, CLI, cron or shift handoff, plus a pre-packaging cleanup checklist.

Price: 3,499,000 VND
Level: Intermediate
Format: Video, skill-writing labs and shift-handoff templates
Time commitment: About 16 sessions; plan on 6–8 hours per week to write, test-run and clean up skills

## Who it is for

- People who use several AI coding harnesses and want to reuse the same process instead of rewriting prompts
- Operators or developers who need to delegate work via chat, CLI and cron schedules and keep track of shifts
- Small teams that want to turn manual processes into skills that can be tested, shared and maintained

## Before you start

- Basic comfort with the terminal, Markdown files and an AI coding assistant
- Understand environment variables, API keys and the rule of keeping secrets out of source code
- Have a real process of your own to turn into a skill and test

## Curriculum

### Module 1 — Where Hermes Fits (2 sessions)

Outcome: Hermes is running, and you understand how it differs from a regular agent.

- The operations layer vs. the agent layer: where responsibilities end
- What carries over between harnesses and what doesn't
- Installing Hermes and running your first shift
- How to delegate so you get usable results back

### Module 2 — Skills Are the Unit of Reuse (3 sessions)

Outcome: 3 skills of your own, running across multiple harnesses.

- Anatomy of a skill: inputs, process, outputs
- The 6-step process: from a task you do by hand to a skill
- Importing a skill into Hermes and checking that it runs correctly
- Naming and organizing so you can still find it 6 months later

### Module 3 — Cleaning Up Before You Package (2 sessions)

Outcome: Clean skills you can share without leaking anything.

- Checklist: tokens, API keys, machine paths, private data
- Separating configuration from skill content
- Why you should never package a raw export
- Self-reviewing a skill before handing it to anyone else

### Module 4 — Four Ways to Run (3 sessions)

Outcome: One skill that runs all 4 ways without being rewritten.

- Chat: tasks that need back-and-forth discussion
- CLI: one-off calls that return a result right away
- Cron: recurring tasks, choosing the right cadence
- Shift handoff: keeping context between runs

### Module 5 — Connecting to Real Tools (3 sessions)

Outcome: Hermes can read from and write to the systems you already use.

- MCP: letting Hermes use external tools
- Connecting your database and APIs with limited permissions
- Chaining several skills into one long workflow
- Placing approval checkpoints at steps that could cause damage

### Module 6 — A Shift That Runs Every Day (3 sessions)

Outcome: A set of operating skills and a shift schedule that are actually running.

- Deciding which tasks to delegate and which to keep doing yourself
- The shift schedule: daily and weekly tasks
- Reading logs to tell whether the shift did the work right
- Distilling repeated work into new skills and growing your asset library

## What you get

- 6 modules, ending with 1 operating skill set of your own
- Write a skill once, run it on multiple harnesses
- Four run modes: chat, CLI, scheduled cron, shift handoff
- Cleanup checklist: no tokens, keys or private data slip through
- A 6-step process for turning manual work into a reusable skill
- Skill templates + direct support in the VIP Discord

## Details

Hermes isn't another chatbot. It's a duty shift: you hand off the work, it runs it with standardized skills, leaves a log and hands over to the next run.

Its real strength is reuse. A skill written to the standard runs on Claude Code, Codex, Cursor and Gemini — with no rewriting for each one.

You'll learn:
• Writing skills: turning work you currently do by hand into a reusable asset
• Four ways to run: chat, CLI, scheduled cron, and handoffs between shifts
• Cleaning up before packaging: no tokens, keys or private data slipping through

The rule that runs through the whole course: a skill has to be clean before it can be packaged. No tokens, no keys, no raw exports, no personal data.

## Frequently asked questions

### I'm new to AI coding assistants. Can I keep up?

You should be comfortable with at least one tool such as Codex, Claude Code, Cursor or Gemini, and know the terminal basics. The course focuses on skill design and operations; it doesn't re-teach programming fundamentals.

### What process do I need to prepare for the hands-on practice?

Bring one repetitive task with clear inputs and outputs, such as compiling a report or reviewing content. Don't use personal data, tokens or raw exports in the exercises.

### After finishing the course, can Hermes run things for me completely?

You can build a set of skills and shifts for suitable tasks, but you still need to monitor the logs and approve the risky steps. Results depend on your process, tools and testing quality; the course does not guarantee income.

### How much support do I get?

The course includes skill templates, multi-harness examples and VIP Discord support for the labs. Support helps you track down design or run-mode errors; it does not accept system secrets and does not operate your production environment for you.

### Which agents do the skills in the course run with?

Skills are written to the reuse standard and run on Claude Code from Claude, Codex from ChatGPT, Cursor and Gemini, as described in the course introduction. You use your own account for each tool.
