folder_openGetting Started schedule5 min read update3 Nov 2026

This is a template. It ships with the folder structure, index notes, plugin manifests, and scripts for a shared Obsidian “second brain” that Hermes writes session archives and staged memory candidates into — organized using the PARA Method (01-Projects/, 02-Areas/, 03-Resources/, 04-Archives/, _System/). It contains no personal data, no chat history, and no API keys. Follow these steps to stand up your own copy.

You can install Hermes Brain directly with a single terminal command from GitHub, or by running the installer inside a cloned template repository.

Option 1: One-Line Install from GitHub (No manual clone needed)

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

Option 2: From Cloned Repo or Local Template

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

Both installers:

  • copy/clone the template to the destination you choose (default ~/Hermes Brain)
  • optionally set up Ollama or sentence-transformers for the trend-digest scripts
  • detect Obsidian and tell you which plugins to enable if it’s missing anything
  • detect the hermes CLI and register the vault path as env.HERMES_VAULT_PATH
  • offer to register the hourly session-archiving cron job (hermes-brain-archive-hourly)
  • optionally run consolidate_memory.py once as a sanity check

If you’d rather do it by hand, or the installer can’t run in your environment, follow the manual steps below.

1. Place the vault (manual path)

Copy this whole folder anywhere on your machine, e.g.:

  C:\Users\<you>\<wherever>\Hermes Brain

It works fine inside OneDrive/Dropbox/Google Drive for sync, or fully local — your call. There’s nothing that assumes a specific location; scripts resolve their own paths relative to _System/Scripts/.

2. Open it in Obsidian

  • Install Obsidian if you don’t have it.
  • “Open folder as vault” → select this folder.
  • Obsidian will ask to trust/enable the bundled community plugins. Enable:
    • Dataview — powers the live tables in 04-Archives/Daily/Timeline.md and Dashboards.
    • Smart Connections — semantic search/related-notes across the vault.
    • Local REST API — lets Hermes (or scripts) write/read notes over HTTP.
    • Kanban — powers 04-Archives/Memory-Review/Memory-Board.kanban.
  • If Obsidian says a plugin is missing/disabled, go to Settings → Community Plugins and toggle it on manually.
  • Minimum plugin versions are documented in .obsidian/plugin-versions.json.

3. Generate your own Local REST API key

The template ships without any API key.

  • Settings → Local REST API → it auto-generates a key on first enable.
  • Store it in Hermes’s protected env, not in a vault note:
      hermes config set env.OBSIDIAN_REST_API_KEY "<your-key>"
  • Never paste it into a .md file in this vault.

4. Point Hermes’s session archiver at this vault

Hermes archives chat sessions into 04-Archives/Daily/YYYY/MM/DD/ using either:

  • Real-Time Lifecycle Hooks (instant, recommended): Add shell hooks to ~/.hermes/config.yaml (on_session_start, post_llm_call, on_session_end) pointing to _System/Scripts/hermes_session_sync.cmd (or .sh). New chats create notes instantly; existing chats update on every turn. See INSTALL_PROMPT.md for the YAML snippet.
  • Hourly Cron (background catch-up): An hourly cron job (0 * * * *) that executes python _System/Scripts/hourly_archive.py. Uses database-aware detection so active sessions are never missed.

See Hermes-Agent-Vault-Setup for the pipeline this vault expects: 04-Archives/Daily → 04-Archives/Memory-Review → native MEMORY.md/USER.md.

5. Nightly memory consolidation & dream cycle

_System/Scripts/consolidate_memory.py and _System/Scripts/dream_cycle.py deduplicate, stage candidates, and synthesize session insights:

  • consolidate_memory.py: Mechanical, LLM-free extraction of facts into 04-Archives/Memory-Review/.
  • dream_cycle.py: LLM-assisted noctural synthesis summarizing key takeaways into 03-Resources/Research/.
  • sync_to_hermes.py: Syncs promoted facts to Hermes native memory.

Run a sanity check:

  python _System/Scripts/consolidate_memory.py
python _System/Scripts/sync_to_hermes.py --dry-run

6. Semantic search layer

_System/Scripts/semantic_search.py builds a local vector search index over the vault’s session and knowledge files using sentence-transformers/all-MiniLM-L6-v2 (22 MB, CPU-only, no external API). Supports incremental updates, CLI search, and an HTTP server for real-time DataviewJS queries.

  # Install deps first
pip install sentence-transformers numpy

# Build the index (full rebuild)
python _System/Scripts/semantic_search.py --build

# Incremental update (run periodically)
python _System/Scripts/semantic_search.py --update

# Test a search query
python _System/Scripts/semantic_search.py --search "prompt optimization" --top 5

# Run HTTP server for Dashboard widget (default port 8765)
python _System/Scripts/semantic_search.py --serve

The Dashboard’s semantic search widget will query http://localhost:8765/search?q=... when the server is running.

7. Skill forecasting

_System/Scripts/skill_forecast.py analyzes vault activity trends and installed skills to suggest which skills to learn/install next with recency-weighted trends and gap analysis.

  python _System/Scripts/skill_forecast.py --days 30 --top 10

The script reads from 04-Archives/Audit-Reports/Tag-Trends.log, 02-Areas/Skills/Installed-Skills-Index.md, and 02-Areas/Skills/Skill-to-Chat-Links.md. External mappings are managed in _System/Scripts/tag_skill_map.json.

8. Hourly session archiving + token tracking

_System/Scripts/hourly_archive.py is the core automation that:

  • Exports sessions into 04-Archives/Daily/YYYY/MM/DD/
  • Maintains a deduplicated 04-Archives/Daily/manifest.jsonl
  • Extracts token usage and appends daily totals to 04-Archives/Audit-Reports/Token-Usage.log

Run it once manually to confirm:

  python _System/Scripts/hourly_archive.py

Schedule it hourly via Task Scheduler, cron, or Hermes’s cronjob_manage tool.

9. Vault audit & agent performance

_System/Scripts/vault_audit.py scans the vault for broken wikilinks, stale Memory-Review tasks, and orphans, writing its report to 04-Archives/Audit-Reports/Vault-Audit-Report.md. _System/Scripts/agent_performance.py summarizes skill and model usage trends into 04-Archives/Audit-Reports/Agent-Performance.md.

  python _System/Scripts/vault_audit.py
python _System/Scripts/agent_performance.py

10. Pulling template updates

_System/Scripts/Installers/update.sh (Linux/macOS) and update.ps1 (Windows) turn the vault into a local git repo tracking the template remote to fetch and merge upstream template updates safely.

  .\_System\Scripts\Installers\update.ps1

Personal content (04-Archives/Daily/, 04-Archives/Memory-Review/*, 01-Projects/* beyond starter templates) is excluded by .gitignore and protected from clobbering.

What’s NOT included (by design)

  • Any chat history, session transcripts, or personal data
  • Any API keys, credentials, or Local REST API keys
  • Generated caches (.smart-env/, cache/graphify/, 04-Archives/Audit-Reports/*.log, 04-Archives/Audit-Reports/*.md)
  • Anyone’s hardcoded personal machine paths