Obsidian Just 10x’d Everyone’s Hermes Agent
By David Ondrej
Key Concepts
- Second Brain: A digital system (Obsidian) used to store, organize, and retrieve information, goals, and business data.
- Living Files: Markdown files that are accessible, readable, and editable by AI agents, allowing them to be used as context, memory, or skills.
- Hermes Agent: An autonomous AI agent capable of executing tasks, managing files, and utilizing skills.
- VPS (Virtual Private Server): A dedicated cloud-based computer used to run AI agents 24/7, independent of the user's local machine.
- Context Engineering: The practice of organizing data (notes/skills) so that AI models can access relevant information on-demand without bloating the token window.
- Tmux: A terminal multiplexer used to manage multiple terminal sessions (e.g., running an AI agent in one pane and monitoring the VPS in another).
- Obsidian Sync: An encrypted service to synchronize vaults across multiple devices (MacBook, phone, VPS).
1. The "Living File" Philosophy
The author argues that most digital files are "dead" because AI agents cannot access, compare, or research them. To make a file "living," it must be stored in a format (Markdown) that an AI can use as a skill, memory, or context. By integrating Obsidian with an AI agent, the user transforms their notes into a dynamic, machine-readable knowledge base.
2. Technical Architecture & Setup
- VPS Infrastructure: The author recommends using Hostinger to host a KVM 2 VPS. This provides a dedicated environment for agents like Hermes, Pi, or CodeX to run continuously.
- Terminal Workflow: The author utilizes iTerm and Tmux to manage the VPS. This allows for a split-screen workflow where one pane runs an "orchestrator" agent (like Pi Agent) to manage the server, while the other executes tasks.
- Obsidian Integration:
- Syncing: Obsidian Sync is used to keep the vault consistent across devices.
- Headless Obsidian: The author uses a headless setup on the VPS to allow the Hermes agent to read and write directly to the vault.
- Memory Management: The system uses a two-tier memory structure:
- Soul.md: A core system prompt (always loaded) containing essential user preferences.
- On-Demand Context: Skills and notes are only pulled into the AI's context window when relevant, preventing token bloat.
3. Step-by-Step Implementation
- Provisioning: Purchase a VPS (Hostinger) and connect via SSH using a terminal (iTerm).
- Agent Setup: Use an orchestrator agent (e.g., Pi Agent) to install Hermes Agent dependencies via plain English commands.
- Configuration: Connect the Hermes Agent to an inference provider (e.g., OpenRouter) and select a model (e.g., GLM 5.2).
- Vault Creation: Create an Obsidian vault and enable Obsidian Sync for cross-device access.
- Headless Sync: Use the provided scripts to link the VPS to the Obsidian vault, enabling the agent to read/write notes autonomously.
4. Key Arguments & Benefits
- Observability: Unlike traditional AI agents that work in "invisible" silos, this setup allows the human to see exactly what the agent is doing via the Obsidian graph and file structure.
- Compounding Knowledge: By saving research and outputs into Markdown files, the user avoids redundant work and token waste.
- Automation Leverage: The
/gofeature in Hermes Agent allows for complex, multi-step goals (e.g., "scrape 50 videos and save transcripts") that run autonomously until the specific, quantifiable outcome is achieved. - Human-Agent Collaboration: The author emphasizes that the user does not need to be a DevOps expert; they only need to be an "orchestrator" who defines goals in plain English.
5. Notable Quotes
- "If there is no easy way for a powerful AI agent to access them, those files are useless." (On the concept of "dead files").
- "Having the right Agentic setup is really the biggest productivity cheat code in 2026."
- "Stop trying to do everything manually. Just sit down for 30 minutes, create a nice proper setup that lets you do your work more efficiently."
6. Synthesis/Conclusion
The integration of Obsidian and Hermes Agent on a VPS represents a shift from passive note-taking to active, agent-driven knowledge management. By treating notes as "living files," users can outsource repetitive research and organizational tasks to an AI that operates 24/7. The primary takeaway is that the combination of visual observability (Obsidian) and autonomous execution (Hermes/VPS) provides a significant competitive advantage in productivity and context management.
Chat with this Video
AI-PoweredLoad the transcript when you're ready to chat so the initial page stays lighter.
Related Videos

User Signal Dies at the Retrieval Boundary - Sonam Pankaj, StarlightSearch
AI Engineer

HTML is All You Need (for Agents to Make Graphics) - Amol Kapoor, Nori
AI Engineer

This AI Brain Will Make You So Smart It’s Almost Unfair
Dan Martell

Stanford MS&E435 Economics of the AI Supercycle | Spring 2026 | Applications, Coding AI
Stanford Online

A Genius With Amnesia - Victor Savkin, Nx
AI Engineer

I made my SaaS ready for AI agents (in San Francisco)
Marc Lou

Implementing DeepMind innovation: Deep research API
Google Cloud Tech