Hermes Brain Doc

Overview

Tue, 03 Nov 2026 00:00:00 +0530

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

Hermes Brain is an Obsidian vault template. Every session your agent runs is archived as plain Markdown, deduplicated, secret-scrubbed automatically, and staged for you to review before anything becomes permanent. There is no black box and no vendor lock-in. It is just a vault, so you can search it, link it, graph it, and back it up like any other notes.

Quickstart

Tue, 03 Nov 2026 00:00:00 +0530

Get Hermes Brain running on your machine in under a minute.

Option 1 — One-line install (fastest)

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

The installer:

Clones the template to your destination (default ~/Hermes Brain)
Detects Obsidian and tells you which plugins to enable
Detects the hermes CLI and registers env.HERMES_VAULT_PATH
Offers to register the hourly session-archiving cron job

Option 2 — GitHub “Use this template”

Click Use this template to create your own private repo.
Clone it and run the installer from inside the folder:

 ./_System/Scripts/Installers/install.sh # Linux/macOS

Step 2 — Open in Obsidian

Install Obsidian (free).
“Open folder as vault” → select your Hermes Brain folder.
Enable the bundled community plugins when prompted:
- Dataview — live tables in the Daily Timeline and Dashboards
- Smart Connections — semantic search across the vault
- Local REST API — lets Hermes read/write notes over HTTP
- Kanban — powers the Memory Review board

Step 3 — Generate your Local REST API key

The template ships without any API key.

Installation Options

Tue, 03 Nov 2026 00:00:00 +0530

🌐 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:

Tutorial: Your First Memory

Tue, 03 Nov 2026 00:00:00 +0530

In this tutorial you will follow a single fact through the whole pipeline: archive, stage, review, and remember. It takes about ten minutes and you need nothing beyond a working install.

info

Start from the vault root (the folder that contains 01-Projects/ and _System/). Use python on Windows and python3 on Linux or macOS. If you have not installed yet, do the Quickstart first.

What you will do

Step	Stage	Where the fact lives afterwards
1. Check the install	n/a	n/a
2. Archive a session	Archive	`04-Archives/Daily/YYYY/MM/DD/`
3. Stage candidates	Stage	`04-Archives/Memory-Review/`
4. Review them	Review	`MEMORY.md`, `02-Areas/User-Profile.md`, or dismissed
5. Sync to Hermes	Remember	Hermes native memory

1. Check the install

Run the MCP server’s self-test and list pending candidates. Neither needs Hermes to be running.

Setup Guide

Tue, 03 Nov 2026 00:00:00 +0530

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.

Quick install (recommended)

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

Install Prompts

Tue, 03 Nov 2026 00:00:00 +0530

Copy-paste blocks below straight into a Hermes/OpenClaw Agent chat. No manual cloning, no running scripts by hand — Hermes/OpenClaw does it for you and asks the same questions the installer would (where to put the vault, which embedding backend, etc).

[!TIP] Prefer a terminal one-liner instead of chatting?

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

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

1. One-shot install prompt

Paste this into Hermes:

Architecture

Tue, 03 Nov 2026 00:00:00 +0530

Hermes Brain turns raw chat sessions into durable, human-reviewed agent memory through four stages.

The four stages. Only stage 3 needs you, and nothing skips it.

Stage 1 — Archiving

Raw agent sessions are exported to 04-Archives/Daily/YYYY/MM/DD/ as plain Markdown. A manifest.jsonl tracks every exported session for deduplication.

Vault Structure

Tue, 03 Nov 2026 00:00:00 +0530

The vault follows the PARA Method (Projects, Areas, Resources, Archives) with a dedicated _System/ core.

Folder	What lives there
`01-Projects/`	Active projects with goals and deadlines, architecture decision records (`01-Projects/ADR/`), and a Dataview database (`Projects.base`)
`02-Areas/`	Long-term standards, human alignment parameters (`02-Areas/User-Profile.md`), the installed-skills catalogue, the team roster, and the 11-dimension Personality Judgment Framework
`03-Resources/`	Domain knowledge hubs (Technology Stack, Cybersecurity, AI/ML, Agentic Architecture MOCs), research inquiries, and reference guides
`04-Archives/`	Daily session timeline, chat correlation, the memory-staging Kanban board, and telemetry and audit reports
`_System/`	Reusable note templates, cross-platform automation scripts, and visual canvases
`Dashboard.md`	A live command centre: active projects, token telemetry, archiver health, recent sessions, and skill analytics

Preconfigured Obsidian plugins: Dataview, Smart Connections, Local REST API, and Kanban, plus two bundled themes, Tokyo Night and Nemoclaw.

Note Templates

Tue, 03 Nov 2026 00:00:00 +0530

Central documentation and reference guide for all Obsidian note templates powering the Hermes Brain vault.

[!TIP] Interactive Template Hubs

Live Dataview Review: Open Templates-Review.md for a Daily-Review style dynamic audit table and verification checklist.

Raw Source Gallery: Open Template-Showcase.md for 1-click copyable raw markdown code blocks.

Core Templates

7 SCHEMAS

Template Engine

OBSIDIAN CORE

Daily Review Link

AUTOMATED

Layout Standard

RESPONSIVE CARDS

🧭 Template Quick Reference Matrix

Template	File	Target Destination	Frontmatter `type`	Primary Persona / System
Project Dashboard	`Project.md`	`01-Projects/`	`project`	Codex (Lead), Argus, Ledger, Vox
Architecture Decision Record	`Architecture-Decision-Record.md`	`01-Projects/ADR/`	`adr`	4-Agent Consensus Review
Lesson Learned	`Lesson-Learned.md`	`02-Areas/Skills/Lessons-Learned/`	`lesson-learned`	Argus (Safety & Risk), Self-Correction
Daily Review	`Daily-Review.md`	`04-Archives/Daily/YYYY/MM/`	`daily-review`	Daily Notes Plugin, Archiver Dataview
Research Note	`Research-Note.md`	`03-Resources/Research/`	`research`	Deep Research, Literature & Tech Spikes
Personality Judgment Analysis	`Personality-Judgment-Analysis.md`	`02-Areas/` or `04-Archives/Daily/`	`analysis`	Vox (Human Alignment & Cognition)
Personality Judgment Dashboard	`Personality-Judgment-Dashboard.md`	`02-Areas/` or `04-Archives/Daily/`	Dynamic	Dataview & DataviewJS Telemetry
Memory Review Candidate	`04-Archives/Memory-Review/TEMPLATE.md`	`04-Archives/Memory-Review/`	Candidate	Memory Review Kanban Board

📖 Comprehensive Template Specifications

1. Project Dashboard

File: _System/Templates/Project.md
Target Location: 01-Projects/<Project-Name>.md
When to Use: Starting an active engineering initiative, complex agent workflow, or cross-tool integration with specific milestones, deliverables, and team assignments.

YAML Frontmatter:

 ---
type: project
status: active # active | completed | paused | on-hold
created: {{date}}
tags:
 - project
cssclasses:
 - dashboard-beta
 - dashboard-wide
---

Structural Architecture:
- Meta Banner: Displays status badge, multi-agent squad pill, and repository/workspace paths.
- KPI Strip: 4 status metrics (Status, Priority, Lead Agent, Linked ADRs).
- Column 1 (Strategy & Squad): Goal & Executive Scope, Repository & Workspace Path, Architecture & Tech Stack, Multi-Agent Squad & Role Matrix (Codex, Argus, Ledger, Vox).
- Column 2 (Execution & Decisions): Operational Stage, Action Checklist, Linked ADRs (ADR Index), and References.
Obsidian Placeholders: {{title}}, {{date}}.

2. Architecture Decision Record (ADR)

File: _System/Templates/Architecture-Decision-Record.md
Target Location: 01-Projects/ADR/ADR-<Number>-<Title>.md
When to Use: Deciding significant technical, architectural, or structural changes that carry trade-offs or constraints.

YAML Frontmatter:

 ---
type: adr
status: proposed # proposed | accepted | rejected | superseded
date: {{date}}
deciders: [Human, Codex, Argus, Ledger, Vox]
tags: [adr, architecture, multi-agent]
---

Structural Architecture:
- Context & Problem Statement: What technical problem or question are we solving?
- Multi-Agent Persona Review:
  - 🛡️ Argus (Security & Risk): Threat modeling, secret handling, attack surface.
  - ⚡ Codex (Implementation & Performance): Code complexity, maintainability, latency.
  - 📊 Ledger (Resource Efficiency & Cost): Token consumption, model inference costs, storage.
  - 🗣️ Vox (Human Alignment & UX): Developer ergonomics, communication clarity.
- Decision & Consensus: Final chosen solution and rationale.
- Consequences: Positive gains vs. accepted trade-offs.
- Footer: Links to Architecture Decision Records Index.

3. Lesson Learned

File: _System/Templates/Lesson-Learned.md
Target Location: 02-Areas/Skills/Lessons-Learned/LL-<Topic>.md
When to Use: Post-incident review following an agent hallucination, API failure, tool misconfiguration, or bad assumption to prevent recurrence.

YAML Frontmatter:

 ---
type: lesson-learned
date: {{date}}
agent_profile: default
status: active
severity: low # low | medium | high
tags: [lesson-learned, self-correction, hermes-brain]
---

Structural Architecture:
- Metadata Banner: Incident date, agent profile, severity level.
- 1. What Happened: Trigger, prompt context, and erroneous action.
- 2. Root Cause Analysis: Why the mistake happened (outdated docs, hallucinated flags, path bugs).
- 3. Human Correction & Fix: How the issue was resolved.
- 4. 🛡️ Standing Preventative Directive: Strict, actionable directive the agent must adhere to in all future sessions.
- Footer: Links to Lessons Learned Catalog.

4. Daily Review

File: _System/Templates/Daily-Review.md
Target Location: 04-Archives/Daily/YYYY/MM/DD.md (or YYYY-MM-DD.md)
When to Use: Daily reflection and session triage. Automatically used by Obsidian’s Daily Notes plugin.

YAML Frontmatter:

 ---
type: daily-review
status: draft
created: {{date}}
tags: [daily-review]
---

Dynamic Dataview Integration: Lists all Hermes chat session archives exported on this day from 04-Archives/Daily/YYYY/MM/DD/:

 TABLE file.mtime AS "Modified", file.size AS "Size"
FROM "04-Archives/Daily" OR "Daily"
WHERE (
 file.folder = this.file.folder + "/" + this.file.name
 OR file.folder = this.file.folder
 OR contains(file.folder, replace(this.file.name, "-", "/"))
)
AND file.name != this.file.name
AND file.name != "README"
AND file.name != "manifest"
AND file.name != "Timeline"
AND file.name != "Chat-Correlation"
SORT file.name ASC

Reflection Checklists:
- Sessions reviewed
- Durable facts found (candidates for 04-Archives/Memory-Review/)
- Notes & follow-ups
- Link to Memory-Review Promotion Criteria.

5. Research Note

File: _System/Templates/Research-Note.md
Target Location: 03-Resources/Research/<Topic>.md
When to Use: Deep dives into technical topics, libraries, prompt engineering patterns, security audits, or agent architectures.

YAML Frontmatter:

 ---
type: research
status: in-progress # in-progress | complete | abandoned
created: {{date}}
tags: [research]
---

Structural Architecture:
- Question / Goal: The explicit question being investigated.
- Findings: Key discoveries and takeaways.
- Sources: Papers, GitHub repos, official docs, benchmarks.
- Open Threads: Remaining unanswered questions.
- Related: Reciprocal wikilinks to relevant vault notes and Research Hub.

6. Personality Judgment Analysis

File: _System/Templates/Personality-Judgment-Analysis.md
Target Location: 02-Areas/Skills/ or 04-Archives/Daily/
When to Use: In-depth cognitive, communicative, and psychological alignment analysis of conversational subjects or human-agent interaction patterns.

YAML Frontmatter:

 ---
type: analysis
status: active
created: {{date}}
tags: [analysis, personality-judgment]
subject: ""
period: ""
primary_style: ""
top_hypothesis: ""
confidence: ""
hypotheses: []
anomalies: []
recommendations: []
topic_behavior: []
---

12 Analytical Dimensions:
1. Subject Profile (metadata, platform, data quality)
2. Communication Style & shifts (formal, technical, defensive, casual, etc.)
3. Language Pattern (vocabulary, sentence styles, grammatical voice)
4. Repetition Analysis (event, complaint, justification, emotional functions)
5. Response Behavior (work, conflict, casual, emotional)
6. Conversation Dynamics (initiation, topic control, question behavior)
7. Conflict & Disagreement (precursors, behaviors, repair mechanisms)
8. Topic Behavior (message counts, latency, emotional variance)
9. Baseline vs. Change (deviation from normal baseline)
10. Anomaly Detection Log (tone, latency, volume anomalies)
11. Context Node Summary (temporal, relational, developmental, environmental)
12. Hypothesis Layer with Confidence Justification (High/Med/Low percentage rationale)
Framework Link: Personality Judgment Framework.

7. Personality Judgment Dashboard

File: _System/Templates/Personality-Judgment-Dashboard.md
Target Location: Dedicated hub note in 02-Areas/ or 04-Archives/
When to Use: Instantiating a real-time command center summarizing all active personality judgment analyses in the vault.
Key Widgets & DataviewJS:
- Active Analyses Table: Lists all subjects, styles, top hypotheses, and confidence.
- Confidence Distribution: Dynamic DataviewJS counts of High, Medium, Low, and Very Low findings.
- Recent Hypotheses: Flattened array query showing the latest 10 hypotheses.
- Timeline: Historical chronology of evaluations.
- Topics Analyzed: Aggregated frequency bar of recurring behavioral topics.
- Active Anomalies: Surface flagged behavior spikes across all notes.

Bonus: Memory Review Candidate Template

File: 04-Archives/Memory-Review/TEMPLATE.md
Target Location: 04-Archives/Memory-Review/<Candidate-Topic>.md
When to Use: Staging a candidate fact extracted from a daily chat transcript before manually approving it into Hermes native MEMORY.md or USER.md.
4 Gatekeeper Criteria:
- Durable (matters in 6+ months)
- Non-sensitive (zero API keys, credentials, or PII)
- Verified (cited from session transcripts or repeatable observation)
- Actionable (changes future agent behavior)
Verdict Checkbox: Promote to memory · Promote to user profile · Archive.

🚀 How to Use Templates in Obsidian

Method 1: Core Templates Plugin (`Ctrl+T` / `Cmd+T`)

Create a new note anywhere in your vault (Ctrl+N / Cmd+N).
Press Ctrl+T (Windows/Linux) or Cmd+T (macOS), or open the Command Palette (Ctrl+P / Cmd+P) and type Insert template.
Select any template from the list (e.g. Project, Architecture-Decision-Record, Lesson-Learned).
Obsidian will automatically populate the note and expand {{date}}, {{time}}, and {{title}}.

Method 2: Daily Notes Button (One Click)

Click the “Open today’s daily note” icon in Obsidian’s left ribbon (or press your daily note shortcut).
Obsidian creates 04-Archives/Daily/YYYY/MM/DD.md using Daily-Review.md.
The note immediately renders the live Dataview table of all chat sessions recorded today.

Method 3: Via Hermes Agent Prompt

You can instruct your Hermes agent to create a note adhering to any template:

Graph View

Tue, 03 Nov 2026 00:00:00 +0530

Graph View Enhancements

The template includes optimized Obsidian graph view settings and CSS enhancements for better visualization:

Features:

Color-coded folders: Standard template folders (01-Projects, 02-Areas, etc.) are automatically colored
Improved node labeling: Clearer, more readable node labels with subtle background
Enhanced link visualization: Subtle but visible connections that highlight on hover
Better node interaction: Nodes enlarge slightly on hover for easier selection
Optimized physics: Adjusted forces for less cluttered, more organic layouts
Depth perception: Slight text fading for distant nodes
Orphan highlighting: Dashed borders on orphan nodes for easy identification

To Enable:

The graph-enhancements.css snippet is already installed in .obsidian/snippets/
Enable it via: Settings → Appearance → CSS snippets
The graph settings in .obsidian/graph.json are pre-configured for optimal viewing

Standard Folder Colors:

🔵 01-Projects: Blue (#4facfe)
🌸 02-Areas: Pink (#ff9ff3)
🌟 03-Resources: Yellow (#feca57)
🔴 04-Archives: Red (#ff6b6b)
💙 Templates: Cyan (#48dbfb)
🍊 00-Inbox: Orange (#ff9f43)
💜 Skills: Purple (#bb86fc)
🟢 Daily Notes: Teal (#00b894)

Note: These colors are assigned based on folder names/tags. Notes not in these folders will use default theme colors.

Agent Skills

Tue, 03 Nov 2026 00:00:00 +0530

This repository contains 7 multi-harness agent skills for the Hermes Brain Obsidian vault template. Each skill is available as a GitHub Agent in .github/agents/ and as a standalone Python script in _System/Scripts/.

Agent Skills

Agent	File	Purpose
self-critique	`_System/Scripts/self_critique.py`	Vault health audit — broken wikilinks, missing frontmatter, module-level paths, hardcoded paths, secret exposure
auto-remediate	`_System/Scripts/auto_remediate.py`	Apply safe fixes from Self-Critique findings (missing `--vault`, missing frontmatter, broken links)
knowledge-synthesis	`_System/Scripts/knowledge_synthesis.py`	BASB/ZK knowledge consolidation — atomic ideas, progressive summarization, MOCs, backlinks
dream-cycle	`_System/Scripts/dream_cycle.py`	Cognitive reflection — cross-session associations, drift detection, ADR drafting, memory staging
quiz	`_System/Scripts/quiz.py`	Self-testing from vault notes — what is X?, complete the statement, true/false
radar	`_System/Scripts/radar.py`	Standing surveillance on sources (blogs, news, papers) → daily/weekly digest
metrics	`_System/Scripts/metrics.py`	Dated vault metrics snapshot — notes, links, sessions, tags, MOCs, stale
converse-learn	`_System/Scripts/converse_learn.py`	Conversation learning — extract learnings from sessions across 4 categories
zknote	`_System/Scripts/zknote.py`	Zettelkasten — fleeting notes, literature notes, MOCs, note splitting, network maintenance

Using Agents

Each agent accepts --vault <path> and runs in --dry-run mode by default. Use --apply to apply changes.

System and Automation

Tue, 03 Nov 2026 00:00:00 +0530

Operational machinery, automation engines, visual canvases, note templates, and internal assets powering the Hermes Brain vault.

Note Schemas

TEMPLATES

Automations

SCRIPTS

Visual Dashboards

CANVASES

📝 Templates & Note Schemas

👉 Documentation: Templates Catalog · Live Review Hub · Raw Gallery

Located in _System/Templates/:

Project Template: Widescreen 2-column project dashboard.
ADR Template: 4-agent persona evaluation framework.
Lesson Learned Template: Post-mortem and prevention schema.
Daily Review Template: Evening reflection and session audit.
Research Note Template: Structured investigation brief.
Personality Judgment Analysis: 7-dimension cognitive evaluation.
Personality Judgment Dashboard: Dataview KPI tracker and cross-session matrix.

💻 Scripts & Automation Engines

Located in _System/Scripts/:

MCP Server

Tue, 03 Nov 2026 00:00:00 +0530

Hermes Brain bundles a Model Context Protocol (MCP) server, _System/Scripts/hermes_brain_mcp.py. It exposes your vault to Hermes Agent, OpenClaw, and other MCP clients such as Claude and Cursor over stdio JSON-RPC.

Tools

Tool	Purpose	Arguments
`vault_search`	Search markdown notes by keywords, tags, or text snippets	`query` (required), `folder` (optional, e.g. `01-Projects`), `limit` (default 10)
`read_note`	Read the full content of a note	`path` (required, relative to the vault root)
`get_user_profile`	Fetch alignment rules, behavioural guidelines, and communication boundaries from `User-Profile.md`	none
`get_active_adrs`	Fetch all accepted Architecture Decision Records so the agent stays consistent	none
`stage_memory_candidate`	Stage a fact into `Memory-Review` for human verification	`fact` (required), `category`, `source`, `rationale`
`vault_stats`	Vault telemetry: session count, active projects, pending reviews, skills	none

stage_memory_candidate accepts these categories: user_preference (default), project_fact, architectural_rule, learned_workflow.

Scripts

Tue, 03 Nov 2026 00:00:00 +0530

Purpose	Windows	Linux/macOS
Memory consolidation (“dream cycle”)	`python _System\Scripts\consolidate_memory.py`	`python3 _System/Scripts/consolidate_memory.py`
Skill-to-Chat-Links regeneration	`python _System\Scripts\generate_skill_links.py`	`python3 _System/Scripts/generate_skill_links.py`
Trend digest — Ollama backend	`_System\Scripts\pipeline.ps1`	`_System/Scripts/pipeline.sh`
Trend digest — sentence-transformers backend	`_System\Scripts\New-TrendDigest.ps1`	`_System/Scripts/trend_digest.sh`
Hourly session archiving + token tracking	`python _System\Scripts\hourly_archive.py`	`python3 _System/Scripts/hourly_archive.py`
Instant session archiving (manual trigger)	`python _System\Scripts\archive_now.py`	`python3 _System/Scripts/archive_now.py`
Session enrichment (add summaries)	`python _System\Scripts\enrich_session.py`	`python3 _System/Scripts/enrich_session.py`
Semantic search index (build/update/serve)	`python _System\Scripts\semantic_search.py`	`python3 _System/Scripts/semantic_search.py`
Session tagging & topic tracking	`python _System\Scripts\session_tagger.py`	`python3 _System/Scripts/session_tagger.py`
Future skill forecast	`python _System\Scripts\skill_forecast.py`	`python3 _System/Scripts/skill_forecast.py`
Vault integrity audit	`python _System\Scripts\vault_audit.py`	`python3 _System/Scripts/vault_audit.py`
Agent performance dashboard	`python _System\Scripts\agent_performance.py`	`python3 _System/Scripts/agent_performance.py`
Install / set up vault + cron	`.\_System\Scripts\Installers\install.ps1`	`./_System/Scripts/Installers/install.sh`
Uninstall / clean up vault	`.\_System\Scripts\Installers\uninstall.ps1`	`./_System/Scripts/Installers/uninstall.sh`
Update template in existing vault	`.\_System\Scripts\Installers\update.ps1`	`./_System/Scripts/Installers/update.sh`
Zero-commitment trial run (scratch vault)	`.\_System\Scripts\Installers\try.ps1`	`./_System/Scripts/Installers/try.sh`
Seed sample/demo data (preview only, opt-in)	`python _System\Scripts\seed_sample_data.py .`	`python3 _System/Scripts/seed_sample_data.py .`

FAQ

Tue, 03 Nov 2026 00:00:00 +0530

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).

Troubleshooting

Tue, 03 Nov 2026 00:00:00 +0530

Start with the built-in checks. They tell you which area to look at.

 # Linux / macOS
./_System/Scripts/Installers/verify_install.sh
python3 _System/Scripts/hermes_brain_mcp.py --test

# Windows PowerShell
.\_System\Scripts\Installers\verify_install.ps1
python _System\Scripts\hermes_brain_mcp.py --test

Installer verification checks

verify_install runs six checks. Here is what each one means and how to fix it.

Check	If it fails
Vault structure (`01-Projects`, `02-Areas`, `03-Resources`, `04-Archives`, `_System`)	You are probably not in the vault root, or the install was interrupted. `cd` into the vault folder and re-run the installer.
Obsidian plugins (`dataview`, `smart-connections`, `kanban` in `.obsidian/plugins/`)	Re-run the installer fully. Then open the vault in Obsidian and enable the community plugins when prompted.
MCP server status (`hermes-agent mcp-server status` should say active)	Start it with `hermes-agent mcp-server start`. This check needs the Hermes CLI in your `PATH`. Without Hermes it will fail, which is expected.
Memory promotion script (`promote_memory.py --list`)	Run it by hand and read the error. Usually Python is missing, or you are in the wrong folder.
`User-Profile.md` has content (`02-Areas/User-Profile.md`, over 50 bytes)	Edit the file and add your details. The installers never overwrite it once it has content.
`.gitignore` ignores `Daily/`	Add a `Daily/` line to `.gitignore` so personal session data is not committed.

The MCP server cannot find my vault

The server looks for the vault in this order:

What's New

Tue, 03 Nov 2026 00:00:00 +0530

Highlights as listed on the live showcase. The full history is on the Changelog page and in the GitHub release notes.

Changelog

Tue, 03 Nov 2026 00:00:00 +0530

All notable changes to the Hermes Brain vault template. Versions correspond to GitHub Releases.

[1.21.0] — 2026-09-18

Added

Future skill forecast widget — Scripts/skill_forecast.py analyzes vault activity trends (from Tag-Trends.log) and installed skills (from Installed-Skills-Index.md) to suggest which skills to learn/install next for maximum relevance. Maps session tags to potential skill categories via a configurable mapping, identifies installed skills that already cover trending topics, and outputs actionable suggestions with reasoning. Runs as a standalone script or can be integrated into cron workflows.
Cross-platform script table updated — added row for skill_forecast.py.

Verified

skill_forecast.py syntax checks clean (python -m py_compile).
Correctly parses 70 installed skills from Installed-Skills-Index.md.
Correctly parses 15 unique trending tags from Tag-Trends.log (top: hermes, linux, llm, automation, obsidian, video).
Tag-to-skill mapping covers all major trending categories.
Output logged to Skills-Notes/Skill-Forecast.log with timestamps.

[1.20.0] — 2026-09-17

Added

Automated session tagging & topic tracking — Scripts/session_tagger.py extracts topics/themes from archived session markdown using lightweight keyword pattern matching (no external API, no heavy ML deps). Runs automatically as part of hourly_archive.py after sessions are moved into YYYY/MM/DD/ directories. Maintains:
- Per-session tags array in Daily/manifest.jsonl
- Global tag cloud with counts in Skills-Notes/Tag-Cloud.log
- Daily tag trends in Skills-Notes/Tag-Trends.log
Dashboard “Tag cloud & trends” widget — renders the tag cloud as styled pills (size/opacity scaled by frequency) with hover effects, plus a 14-day trend history from Tag-Trends.log. Degrades gracefully when source files don’t exist yet.
Cross-platform script table updated — added row for session_tagger.py.

Verified

session_tagger.py syntax checks clean (python -m py_compile).
Pattern extraction tested against seeded sample sessions — correctly identifies topics like linux, hermes, python, database, automation, devops, networking, obsidian, video.
Tag cloud and trends logs written and formatted correctly.
All 13 dataviewjs blocks in Dashboard.md pass node --check; markdown fences balanced.
hourly_archive.py still byte-compiles cleanly after integration.
Scripts/vault_audit.py reports 0 broken links.

[1.19.0] — 2026-09-17

Added

Semantic search layer — 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 full rebuild (--build), incremental update (--update), CLI search (--search), and an HTTP server (--serve) for real-time DataviewJS queries.
Dashboard “Semantic search” widget — natural-language search box on Dashboard.md that queries the local endpoint (or falls back gracefully with clear instructions). Click results to open matching notes.
Cross-platform script table updated — added rows for semantic_search.py, archive_now.py, enrich_session.py.

Verified

semantic_search.py syntax checks clean (python -m py_compile).
Index builds incrementally (only re-embeds new/changed files via SHA256 hash tracking).
Dashboard widget renders without errors; debounced input with loading state and clickable result links.
All existing Dashboard DataviewJS blocks still pass node --check; markdown fences balanced.

[1.18.0] — 2026-09-17

Added

Manifest concurrency lock + instant archive trigger — Scripts/hourly_archive.py (and .ps1) now acquire an exclusive lock file (.hourly_archive.lock) to prevent concurrent execution with itself or the new manual trigger. Added Scripts/archive_now.py (and .ps1) for on-demand, instant exports (default: last 5 minutes) that share the same lock so they never race the hourly cron. Both scripts export redacted markdown + JSONL, update manifest.jsonl, and append to Token-Usage.log.
Optional session enrichment — Scripts/enrich_session.py (and .ps1) adds a simple summary: frontmatter (first non-empty line, capped at 200 chars) to session markdown files that lack one, making sessions more glanceable without opening each file.
Documentation updates — Daily/README.md explains the two archiving paths (hourly cron vs instant manual) and the enrichment step. INSTALL_PROMPT.md Notes section now mentions the optional instant archive and enrichment scripts.

Verified

All scripts (hourly_archive.py, archive_now.py, enrich_session.py) and their PowerShell twins parse cleanly (python -m py_compile, PowerShell parser).
Throwaway-vault test: ran archive_now.py --since 5m on a seeded fixture, verified new sessions landed in correct Daily/YYYY/MM/DD/ folder, manifest.jsonl updated, token usage logged, and no duplicate entries on re‑run.
Enrichment test: ran enrich_session.py on the same fixture, confirmed summary frontmatter added only to files missing it, and existing frontmatter left untouched.
Lock‑file test: launched two instances of hourly_archive.py in parallel — second exited with “Another instance … is already running — exiting.” Same for archive_now.py vs hourly_archive.py.
Scripts/vault_audit.py reports 0 broken links.
Markdown fences in Dashboard.md and README.md remain balanced.

[1.17.0] — 2026-09-16

Added

4 new Dashboard.md sections — the last 4 roadmap items, all shipped together:
- “What’s changed since your last visit” — a delta banner comparing this note’s own frontmatter (written on every open) against current session/candidate counts, so you see “3 new sessions, 2 new candidates since you last opened this” instead of nothing.
- “Storage & file counts” — live file count and total size via app.vault.getFiles(), plus a breakdown of the top 5 file types.
- “Recently promoted” — parses Memory-Review/Consolidation-Log.md (written by consolidate_memory.py) and surfaces the last 5 decided candidates, so the review pipeline feels less like a black box.
- “Top skills used” — parses Skills-Notes/Skill-to-Chat-Links.md (auto-generated by generate_skill_links.py) and ranks the top 5 skills by session count, working against both its tool-call mode and prose-mode output formats.
All 4 sections degrade gracefully when their source file doesn’t exist yet (first-run/no-data messages, same pattern as every other Dashboard section).

Verified: all 11 dataviewjs blocks in Dashboard.md pass node --check; ran a full mock-harness pass with realistic fake data against every block (including a dedicated second-visit test for the delta banner, confirming it correctly detects 3 new sessions and 2 new candidates against saved frontmatter); Scripts/vault_audit.py — 0 broken links; markdown fences balanced (28 fence lines / 14 code blocks).

Security

Tue, 03 Nov 2026 00:00:00 +0530

Supported Versions

We release patches for security vulnerabilities. Which versions are eligible for receiving such patches?

Version	Supported
1.22.x	✅
< 1.22	❌

Reporting a Vulnerability

Please report (suspected) security vulnerabilities to hello@mistrysiddh.com. You will receive a response within 48 hours. If the issue is confirmed, we will release a patch as soon as possible depending on complexity but historically within a few days.

Contributing

Tue, 03 Nov 2026 00:00:00 +0530

Thanks for considering a contribution to this template.

What’s welcome

Fixes to install.sh / install.ps1 for platforms/edge cases they don’t handle yet (different shells, package managers, Obsidian install paths)
New or improved Obsidian plugin recipes documented in SETUP.md
Additional automation scripts (equivalent additions belong in both PowerShell and bash where practical — see Scripts/ for the current pairs)
Documentation clarity fixes

What NOT to contribute

Never open a PR containing real vault content — chat exports, personal Memory-Review entries, populated Research/ notes, .smart-env/ caches, or any .obsidian/plugins/*/data.json. These are git-ignored on purpose; if your local vault has diverged and picked any of these up, check git status before committing. A CI job (Personal Data Guard) also checks this automatically on every push/PR touching those paths and will fail the build if it finds anything disallowed.
API keys, tokens, or credentials of any kind, even example/placeholder ones that look real enough to be reused by mistake.

Before submitting

Test both the Linux/macOS and Windows install paths if you touched install.sh/install.ps1 or anything in Scripts/.
For shell scripts: bash -n script.sh to catch syntax errors.
For PowerShell scripts: run them through the PowerShell parser ([System.Management.Automation.Language.Parser]::ParseFile(...)) or just execute with -WhatIf/dry-run where applicable.
Keep the folder skeleton generic — no hardcoded personal paths (use $PSScriptRoot/$(dirname "${BASH_SOURCE[0]}")-relative resolution, as the existing scripts do).

Questions first?

Use Discussions instead of an issue if you’re not sure something is a bug versus a design choice.

Code of Conduct

Tue, 03 Nov 2026 00:00:00 +0530

Our Pledge

We as members, contributors, and leaders pledge to make participation in our community a harassment-free experience for everyone, regardless of age, body size, visible or invisible disability, ethnicity, sex characteristics, gender identity and expression, level of experience, education, socio-economic status, nationality, personal appearance, race, religion, or sexual identity and orientation.

We pledge to act and interact in ways that contribute to an open, welcoming, diverse, inclusive, and healthy community.

Deployment

Tue, 03 Nov 2026 00:00:00 +0530

This docs site is Hugo with the Lotus Docs theme, built inside Docker and served as static files by nginx.

Traffic path

Browser → host nginx (ports 80/443, TLS) → Docker container (hermes-docs, host port 8085) → nginx inside the container serving the built site.

Project folder

/mnt/Hermes-Brain-Doc on the server.

Docker commands

Task	Command (in `/mnt/Hermes-Brain-Doc`)
Status	`docker compose ps`
Logs	`docker compose logs -f docs`
Stop	`docker compose down`
Start without rebuild	`docker compose up -d`
Rebuild after changes	`docker compose up -d --build`

The container has restart: unless-stopped, so it comes back after a server reboot.

Hermes Brain Doc

Overview

Quickstart

Option 1 — One-line install (fastest) link

Option 2 — GitHub “Use this template” link

Step 2 — Open in Obsidian link

Step 3 — Generate your Local REST API key link

Installation Options

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

Tutorial: Your First Memory

What you will do link

1. Check the install link

Setup Guide

Quick install (recommended) link

Install Prompts

1. One-shot install prompt link

Architecture

Stage 1 — Archiving link

Vault Structure

Note Templates

🧭 Template Quick Reference Matrix link

📖 Comprehensive Template Specifications link

1. Project Dashboard link

2. Architecture Decision Record (ADR) link

3. Lesson Learned link

4. Daily Review link

5. Research Note link

6. Personality Judgment Analysis link

7. Personality Judgment Dashboard link

Bonus: Memory Review Candidate Template link

🚀 How to Use Templates in Obsidian link

Method 1: Core Templates Plugin (Ctrl+T / Cmd+T) link

Method 2: Daily Notes Button (One Click) link

Method 3: Via Hermes Agent Prompt link

Graph View

Graph View Enhancements link

Features: link

To Enable: link

Standard Folder Colors: link

Agent Skills

Agent Skills link

Using Agents link

System and Automation

📝 Templates & Note Schemas link

💻 Scripts & Automation Engines link

MCP Server

Tools link

Scripts

FAQ

Troubleshooting

Installer verification checks link

The MCP server cannot find my vault link

What's New

Changelog

[1.21.0] — 2026-09-18 link

Added link

Verified link

[1.20.0] — 2026-09-17 link

Added link

Verified link

[1.19.0] — 2026-09-17 link

Added link

Verified link

[1.18.0] — 2026-09-17 link

Added link

Verified link

[1.17.0] — 2026-09-16 link

Added link

Security

Supported Versions link

Reporting a Vulnerability link

Contributing

What’s welcome link

What NOT to contribute link

Before submitting link

Questions first? link

Code of Conduct

Our Pledge link

Deployment

Traffic path link

Option 1 — One-line install (fastest)

Option 2 — GitHub “Use this template”

Step 2 — Open in Obsidian

Step 3 — Generate your Local REST API key

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

What you will do

1. Check the install

Quick install (recommended)

1. One-shot install prompt

Stage 1 — Archiving

🧭 Template Quick Reference Matrix

📖 Comprehensive Template Specifications

1. Project Dashboard

2. Architecture Decision Record (ADR)

3. Lesson Learned

4. Daily Review

5. Research Note

6. Personality Judgment Analysis

7. Personality Judgment Dashboard

Bonus: Memory Review Candidate Template

🚀 How to Use Templates in Obsidian

Method 1: Core Templates Plugin (`Ctrl+T` / `Cmd+T`)

Method 2: Daily Notes Button (One Click)

Method 3: Via Hermes Agent Prompt

Graph View Enhancements

Features:

To Enable:

Standard Folder Colors:

Agent Skills

Using Agents

📝 Templates & Note Schemas

💻 Scripts & Automation Engines

Tools

Installer verification checks

The MCP server cannot find my vault

[1.21.0] — 2026-09-18

Added

Verified

[1.20.0] — 2026-09-17

Added

Verified

[1.19.0] — 2026-09-17

Added

Verified

[1.18.0] — 2026-09-17

Added

Verified

[1.17.0] — 2026-09-16

Added

Supported Versions

Reporting a Vulnerability

What’s welcome

What NOT to contribute

Before submitting

Questions first?

Our Pledge

Traffic path

Project folder

Docker commands