Hermes Brain — Obsidian Vault Template for Hermes Agent (including OpenClaw)

Give your Hermes/OpenClaw agent a memory it can't forget — and you can actually read.

Every session your agent runs gets archived as plain markdown, deduped and
secret-scrubbed automatically, and staged for you to review before anything
becomes permanent. No black box, no vendor lock-in — it's just an Obsidian
vault, so you can search it, link it, graph it, and back it up like any
other notes.

🚀 Try it in 60 Seconds (Verifiable)

# 1. Clone the template
git clone https://github.com/mistrysiddh/hermes-brain-template.git my-second-brain
cd my-second-brain

# 2. Run the installer (choose your OS)
#    Linux/macOS:        ./_System/Scripts/Installers/install.sh
#    Windows PowerShell: powershell -ExecutionPolicy Bypass -File .\_System\Scripts\Installers\install.ps1

# 3. Self-test the bundled MCP server (vault_search, read_note, stage_memory_candidate, ...)
python _System/Scripts/hermes_brain_mcp.py --test

# 4. List pending memory candidates (prints [] on a fresh vault)
python _System/Scripts/promote_memory.py --list

✅ Verification: step 3 ends with "All hermes-brain-mcp self-tests PASSED" and step 4 prints []. No Hermes install is needed for either check.

Why this instead of nothing?

• You stop losing context. Every Hermes session gets archived automatically — nothing lives only in a chat log you'll never scroll back to.
• You stay in control of what becomes "memory." Nothing gets promoted to Hermes's real memory without a human reading and approving it first.
• It's just markdown. Open it in Obsidian, grep it, put it in git, read it in Notepad — no proprietary format, no export step.
• Set up once, in one paste. A single prompt into your Hermes chat installs the vault and wires up hourly archiving — see below.

Why this instead of [alternative]?

Use a vector-DB memory service if you want the agent to auto-recall
semantically similar things with zero human-in-the-loop. Use this
template if you want to see and approve what your agent remembers,
in a format you already own.

💡 What You Solve With This Template

What's inside

04-Archives/Daily/YYYY/MM/DD/*.md   →   04-Archives/Memory-Review/*.md   →   Hermes native MEMORY.md / USER.md
     (raw session archive)                    (staged candidate                 (durable, injected every
                                            facts, human-reviewed)            turn — promoted by hand)

The vault is structured according to the PARA Method (Projects, Areas, Resources, Archives) with a dedicated _System/ core:

• 01-Projects/ — Active projects with goals and deadlines, architectural decision records (01-Projects/ADR/), and an Obsidian Dataview database (Projects.base).
• 02-Areas/ — Long-term standards, human alignment parameters (02-Areas/User-Profile.md), installed skills catalog, team roster, and the 11-dimension [[02-Areas/Skills/Personality-Judgment-Framework|Personality Judgment Framework]].
• 03-Resources/ — Domain knowledge hubs ([[03-Resources/MOCs/Technology-Stack-MOC|Technology Stack]], [[03-Resources/MOCs/Cybersecurity-MOC|Cybersecurity]], [[03-Resources/MOCs/AI-ML-MOC|AI/ML]], [[03-Resources/MOCs/Agentic-Architecture-MOC|Agentic Architecture]]), technical research inquiries (03-Resources/Research/), and reference guides (03-Resources/Guides/).
• 04-Archives/ — Historical and completed items: [[04-Archives/Daily/Timeline|Daily Session Timeline]], [[04-Archives/Daily/Chat-Correlation|Chat Correlation]], 3-stage memory staging ([[04-Archives/Memory-Review/Memory-Board.kanban|Memory Kanban]]), and telemetry/audit reports (04-Archives/Audit-Reports/).
• _System/ — Operational core: Reusable note templates (_System/Templates/), cross-platform automation scripts (_System/Scripts/), and visual architecture canvases (_System/Canvases/Memory-Pipeline.canvas).
• Dashboard.md & _System/Canvases/Dashboard-Beta.md — Live command centers providing immediate visibility into active projects, token telemetry, archiver health, recent sessions, and skill analytics.
• Preconfigured Obsidian plugins: Dataview, Smart Connections, Local REST API, Kanban, plus two bundled themes — Tokyo Night and Nemoclaw.

This repo ships as a template only — no personal data, chat history, or
API keys are included. See SETUP.md for the full breakdown of
what was intentionally left out.

Quick start

🌐 Interactive Web Tour: Explore the memory pipeline simulator, live architecture explorer, and MCP tool tester on the Hermes Brain Showcase Website.

🚀 Option 1 — One-Line Terminal Install (Fastest, directly from GitHub)

Open your terminal and run the one-liner. It automatically clones or downloads the latest template directly from GitHub, sets up your vault, configures your Hermes CLI environment (HERMES_VAULT_PATH), and optionally registers the hourly session archiver:

Linux / macOS / WSL:

curl -fsSL https://raw.githubusercontent.com/mistrysiddh/hermes-brain-template/main/_System/Scripts/Installers/install.sh | bash

Windows (PowerShell):

irm https://raw.githubusercontent.com/mistrysiddh/hermes-brain-template/main/_System/Scripts/Installers/install.ps1 | iex

Non-interactive flags are supported: --dest <path> --backend <ollama|python|skip> --test (PowerShell: -Dest <path> -Backend <ollama|python|skip> -Test).

📦 Option 2 — GitHub "Use this template" or Clone

If you want your own private/custom git repository:

1. Click the green Use this template button on GitHub to create your personal vault repository.
2. Clone your repository:

   git clone https://github.com/<your-username>/<your-vault-repo>.git
   cd <your-vault-repo>
   

3. Run the interactive installer from inside the cloned folder:
   • Linux / macOS:

     chmod +x _System/Scripts/Installers/install.sh
     ./_System/Scripts/Installers/install.sh
     

   • Windows (PowerShell):

     powershell -ExecutionPolicy Bypass -File .\_System\Scripts\Installers\install.ps1
     

The installer guides you through destination selection, optional local embeddings (Ollama / sentence-transformers), and registering your vault with the Hermes CLI.

🤖 Option 3 — One Paste into Hermes Agent (Easiest for Hermes users)

If you already run a Hermes agent, skip cloning and manual scripting entirely: open
INSTALL_PROMPT.md and paste the master prompt straight into
your Hermes chat:

• Prompt 1 — installs the vault only
• Prompt 2 — sets up the hourly archiving cron job only (vault must already exist)
• Prompt 3 (master) — does both in one paste: install the vault, then set up the hourly cron job, no follow-up needed

🧪 Option 4 — Zero-Commitment Trial (Scratch Vault)

Not ready to install anything yet? Copy the template into a temporary scratch
directory and open it in Obsidian without touching your real config or
registering anything with Hermes:

Linux / macOS:

./_System/Scripts/Installers/try.sh

Windows (PowerShell):

powershell -ExecutionPolicy Bypass -File .\_System\Scripts\Installers\try.ps1

Delete the scratch copy any time. Run Option 1 or 2 when you are ready to keep it for real.

🛠️ Option 5 — Full Manual Control

Prefer configuring everything by hand? Follow SETUP.md step by step.

🌐 Option 6 — GitHub Codespaces (Zero-Local-Setup Trial)

No local install needed. Click the badge below to spin up a ready-to-run environment in your browser:

Once the Codespace loads, run the scratch-vault trial:

./_System/Scripts/Installers/try.sh

This creates a temporary vault at /tmp/hermes-brain-trial that you can open in Obsidian (via the Obsidian URI scheme or by copying files out). When you're done, just delete the Codespace — no cleanup needed on your machine.

Tip: Codespaces gives you 60 hours/month free on the GitHub Free plan — plenty for evaluating the template.

Cross-platform scripts

Requirements

• Obsidian (free)
• A Hermes Agent install (or OpenClaw), if you want the automated session-archive/memory pipeline
• Optional, for the trend-digest scripts: Ollama or Python 3 with sentence-transformers

FAQ

Do I need to run a local LLM (Ollama, etc.)?
No. The core vault, session archiver, and memory-review pipeline work with any Hermes agent — no local models required. The optional trend-digest scripts (pipeline.sh, trend_digest.py, etc.) are the only parts that need Ollama or sentence-transformers for embeddings.

Is my data private?
Yes. Everything lives in your local vault (or your private fork of this template). No data is sent to this repo or any external service unless you configure it to (e.g., enabling the Local REST API plugin and exposing it).

Can I use this without Hermes / OpenClaw?
Absolutely. The vault structure, templates, Dataview queries, Kanban boards, and scripts are all usable as a standalone Obsidian "second brain" template. The Hermes-specific automation (hourly archiving, INSTALL_PROMPT.md) simply won't run without a Hermes agent.

What's the difference between the "Use this template" button and the one-line install?

• Use this template → creates your own private repo on GitHub with full history. Best if you want to version-control your personal vault.
• One-line install (curl ... | bash) → downloads the latest template directly to your machine, no GitHub account needed. Best for quick local setup.

Why are Daily/, Memory-Review/, and .smart-env/ git-ignored?
They contain your personal session data, API keys, and plugin state. The template only ships the structure (folders, templates, scripts). See SETUP.md for the full list of intentionally excluded paths.

How do I update the template after I've installed it?
Run the update script from inside your vault:

./_System/Scripts/Installers/update.sh     # Linux/macOS
.\\_System\\Scripts\\Installers\\update.ps1   # Windows

This pulls the latest template files (scripts, templates, configs) without touching your personal data in Daily/, Memory-Review/, etc.

Where do I ask questions or request features?
Open a Discussion — there's a welcome post to get started. Bugs and PRs go in Issues.

Security notes

• Never commit real vault content back into this template repo — Daily/,
  Memory-Review/* (except TEMPLATE.md), Research/* (except README.md),
  .smart-env/, and all .obsidian/plugins/*/data.json files are already
  git-ignored for exactly this reason.
• The Local REST API plugin generates its own API key on first enable —
  store it via your Hermes agent's protected env (hermes config set env.OBSIDIAN_REST_API_KEY "..."), never as a plaintext note in the vault.
• Run only one session-archiving cron job per vault — concurrent writers
  to Daily/manifest.jsonl will race and corrupt the index.

Vault Structure

Here is the organized folder structure of the Hermes Brain template:

├── 01-Projects/
│   ├── ADR/
│   │   └── README.md
│   ├── Hermes-Agent-Vault-Setup.md
│   ├── Projects.base
│   └── README.md
├── 02-Areas/
│   ├── README.md
│   ├── Skills/
│   │   ├── Installed-Skills-Index.md
│   │   ├── Lessons-Learned/
│   │   │   └── README.md
│   │   └── Personality-Judgment-Framework.md
│   ├── Team-Profiles-Index.md
│   └── User-Profile.md
├── 03-Resources/
│   ├── README.md
│   ├── Guides/
│   │   ├── Dataview-Query-Library.md
│   │   └── Kanban-Usage.md
│   ├── MOCs/
│   │   ├── AI-ML-MOC.md
│   │   ├── Agentic-Architecture-MOC.md
│   │   ├── Cybersecurity-MOC.md
│   │   └── Technology-Stack-MOC.md
│   └── Research/
│       └── README.md
├── 04-Archives/
│   ├── README.md
│   ├── Daily/
│   │   ├── .gitkeep
│   │   ├── Chat-Correlation.md
│   │   ├── README.md
│   │   └── Timeline.md
│   ├── Memory-Review/
│   │   ├── HERMES-PREAMBLE.md
│   │   ├── Memory-Board.kanban
│   │   └── TEMPLATE.md
│   └── README.md
├── _System/
│   ├── README.md
│   ├── Canvases/
│   │   ├── Dashboard-Beta.md
│   │   ├── Memory-Pipeline.canvas
│   │   └── Personality-Judgment-Canvas.canvas
│   ├── Scripts/
│   │   ├── Installers/
│   │   │   ├── install.ps1
│   │   │   ├── install.sh
│   │   │   ├── try.ps1
│   │   │   ├── try.sh
│   │   │   ├── uninstall.ps1
│   │   │   ├── uninstall.sh
│   │   │   ├── update.ps1
│   │   │   └── update.sh
│   │   ├── agent_performance.py
│   │   ├── archive_now.ps1
│   │   ├── archive_now.py
│   │   ├── build_fixture_vault.py
│   │   ├── consolidate_memory.py
│   │   ├── dream_cycle.py
│   │   ├── enrich_session.ps1
│   │   ├── enrich_session.py
│   │   ├── generate_skill_links.py
│   │   ├── hourly_archive.py
│   │   ├── New-TrendDigest.ps1
│   │   ├── pipeline.ps1
│   │   ├── pipeline.sh
│   │   ├── pull_supermemory.py
│   │   ├── seed_sample_data.py
│   │   ├── semantic_search.py
│   │   ├── session_tagger.py
│   │   ├── skill_forecast.py
│   │   ├── sync_to_hermes.py
│   │   ├── tag_skill_map.json
│   │   ├── trend_digest.py
│   │   └── trend_digest.sh
│   ├── Templates/
│   │   ├── Architecture-Decision-Record.md
│   │   ├── Daily-Review.md
│   │   ├── Lesson-Learned.md
│   │   ├── Personality-Judgment-Analysis.md
│   │   ├── Personality-Judgment-Dashboard.md
│   │   ├── Project.md
│   │   └── Research-Note.md
├── assets/
│   ├── graph-view-screenshot.png
│   └── install-demo.png
├── docs/
│   ├── LICENSE
│   └── CHANGELOG.md
├── .gitignore
├── Dashboard.md
├── INSTALL_PROMPT.md
├── MOC.md
├── README.md
├── SETUP.md
└── Welcome.md

Structure Explanation:

• 01-Projects/ — Active projects with goals and deadlines
• 02-Areas/ — Long-term standards, skills, and user profile
• 03-Resources/ — Domain knowledge (MOCs, research, guides)
• 04-Archives/ — Historical items (daily sessions, memory review)
• _System/ — Operational core (templates, scripts, canvases)

This PARA-based structure keeps everything organized and discoverable.

💖 Support & Sponsor

If you find Hermes Brain useful for your agent workflows and personal knowledge management, consider supporting its maintenance and continued development:

• GitHub Sponsors: sponsor @mistrysiddh
• ⭐ Star this repository: Helping more builders and AI researchers find the project!

✅ Trust Through Verification

Every claim in this README is verifiable by you in <5 minutes:

You run these checks → you trust the template.

License

MIT — see LICENSE. Use, fork, and adapt freely.

Contributing

Issues and PRs welcome — especially more platform-specific install fixes,
additional Obsidian plugin recipes, or extra automation scripts.
Discussions
are enabled if you'd rather ask a question first — start with the
welcome post.
See CODE_OF_CONDUCT.md for community guidelines.