Hermes Skills

A growing collection of reusable Hermes Agent skills, built to be installed, extended, and adapted through natural language — no YAML, no config files, no manual directory wrangling.

By João Silva

---

Installation

Every skill in this repo is designed to be installed the way you would tell a colleague what to do: in natural language.

Each skill lives in its own folder with a README containing a copy-paste install prompt. Send that prompt to your Hermes agent and it handles the rest — cloning the repo, placing the files in ~/.hermes/skills/, and asking you the configuration questions it needs.

Once installed, the skill is available as a slash command (/medium-story, /short-videos, /excalidraw, /ai-projects) — or you can invoke it naturally by describing what you want.

Current Skills

Alternative: Install via Skills Hub CLI (Advanced)

Hermes also has a built-in Skills Hub with CLI commands for power users:

# Add this repo as a permanent tap (external source)
hermes skills tap add ciberjohn/Hermes-Skills

# Install a specific skill from the tap
hermes skills install ciberjohn/Hermes-Skills/medium-story

# Or install directly from a URL
hermes skills install https://raw.githubusercontent.com/ciberjohn/Hermes-Skills/main/medium-story/SKILL.md

# List installed skills
hermes skills list

# Browse available skills from all sources
hermes skills browse

# Search across hub sources
hermes skills search pipeline

Taps add the entire repo as an external skill directory. Skills update automatically when you update the repo.

Natural Language Auto-Discovery

You can also install a skill simply by telling your Hermes agent what you want:

“I need to write a Medium article about why SSH key management still fails in 2026.”

If the medium-story skill is installed, Hermes automatically matches your goal to the skill’s description, loads the full instructions, and runs the pipeline. No slash command needed.

The same applies to every installed skill. The skill descriptions at the top of each SKILL.md file are what Hermes uses for discovery, so the more descriptive they are, the better the agent picks the right one.

---

How Skills Work

Skills are an open standard (compatible with agentskills.io). Each skill is a folder containing a SKILL.md file with metadata and instructions that tell your Hermes agent how to perform a specific task. Skills can also bundle scripts, reference materials, and templates.

Once a skill is in your ~/.hermes/skills/ directory, Hermes discovers it at startup and loads only the name and description — just enough to know when it might be relevant. When you give Hermes a task that matches a skill’s description, it loads the full instructions and executes the pipeline.

You never manage sessions, approve permission prompts, or restart stalled processes. The agent handles everything.

---

The Pattern

These skills follow the same architecture that runs my own fleet: a main agent that delegates work to specialised sub-agents, communicates through messaging gateways, and runs on a schedule through Hermes cron. You do not need that exact setup to use the skills. They work with any Hermes profile.

The key insight is that the skill system lets you package operational knowledge as something an agent can load and follow, the same way a human engineer reads a runbook. The more skills you build, the more your agent understands about your infrastructure, your workflows, and your preferences — all expressed in natural language, all stored as plain markdown files.

---

Repository Structure

Hermes-Skills/
├── skill-writer/             # Meta-skill: creates new skills
├── medium-story/             # Medium article pipeline
├── ciberjohn-no-slop/        # Anti-AI-slop writing constraints + checker
├── short-videos/             # Short video pipeline
├── excalidraw/               # Diagram generation
├── openrouter-image-gen/     # Image generation via OpenRouter API
├── ai-projects/              # Repository sync
├── 3dprinter/                # Flashforge AD5X headless slicing + printing
├── openscad-cad/             # Parametric design: sketch/dims → OpenSCAD → STL
├── t3mp3st-autonomous-security/# Autonomous security ops
├── social-poster/            # Direct OAuth social media posting
├── technical-trainer/        # Course creation pipeline
├── transcribe/               # Local audio transcription pipeline
├── gymcoach/                 # Personal fitness coach skill
├── joplin-brain/             # Self-filing second brain on Joplin
├── uk-business-consultant/   # UK business consultant skill
├── jev/                      # Typed decisions via TypeSafe's Jev (System One)
├── paperclip-creator/        # Self-hosted agent-company platform + Hermes gateway wiring
├── proxmox-estate/           # Proxmox LXC + VM estate: deploy, operate, back up, restore
├── templates/                # Shared templates
│   └── persona-template.md   # Writing voice template
└── scripts/                  # Shared scripts
    └── md_to_html.py         # Markdown → HTML converter

---

Future Skills

This is a living collection. Planned additions include security scanning pipelines, infrastructure health monitors, backup orchestrators, and notification dispatchers — all following the same pattern of natural-language configuration and agent-native execution.

---

License

MIT — use freely, adapt as needed. Attribution appreciated but not required.