r0b0tlab/llm-wiki_obsidian_hermes_r0b0tlabbra1n/llm-wiki-brain
Claude Code and agent skill by r0b0tlab/llm-wiki_obsidian_hermes_r0b0tlabbra1n
- community
Checking that the SKILL.md file is still in the repository.
Filesystem-first LLM-Wiki + Obsidian agent memory system. Use when the agent needs to read, write, search, or maintain its persistent knowledge base. Provides brain search, lint, ingest-sessions, and build-index commands via the r0b0tlabbra1n package.
Install
npx vskill@latest install r0b0tlab/llm-wiki_obsidian_hermes_r0b0tlabbra1n/llm-wiki-brainSKILL.md
LLM Wiki Brain — Hermes Agent Memory System
When to Use
Load this skill when the agent needs to:
- Search its persistent knowledge base for facts, procedures, or project context
- Record new durable information (decisions, failures, successes, project status)
- Check vault health (lint for broken links, stale items, secrets)
- Ingest Hermes sessions into the brain vault
- Navigate the vault structure
Requirements
- Python 3.11+
r0b0tlabbra1npackage installed (pip install r0b0tlabbra1norpip install -e .from the repo)- Brain vault initialized (
brain init --vault PATH) - Hermes
state.dbaccessible for session ingest (read-only)
Quickstart
Core Workflows
1. Orient in a new session
When starting work, an agent should:
brain search "project status" --vault ~/my-brain— get current state- Read
~/my-brain/START_HERE.mdand~/my-brain/_agent/START_HERE.md - Read
~/my-brain/_agent/semantic/project-status.mdfor project context
2. Record a decision
3. Search before acting
Before making changes, search for relevant past context:
4. Lint the vault
Regularly check vault health:
Expected output: zero ERRORs, zero WARNINGs. INFO (orphans) is fine.
5. Ingest new sessions
After significant work, ingest the session:
Run brain build-index --vault ~/my-brain after ingestion.
Vault Structure
Memory Tiers
- L1 HOT: Hermes built-in
MEMORY.md(~2KB) — compact pointers only - L2 WARM: Dashboards, project status, operating rules — auto-loaded
- L3 COLD: Session summaries, detailed pages — retrieved on demand
- L4 ARCHIVE: Raw captures, superseded pages — never auto-loaded
Promotion Rules
- Fact observed 3+ times → promote to L2 WARM
- Fact used 5+ retrievals → promote to L1 HOT as compact pointer
- Page stale > 30 days → demote to L3 COLD
- Page stale > 90 days → demote to L4 ARCHIVE
Secrets Policy
NEVER store secrets, API keys, tokens, or passwords in the vault. Store key names and config paths only. The secret scanner runs on every write.
Common Patterns
Find what you need:
Check vault health:
Update the index after changes:
Bring sessions into the brain:
Pitfalls
- Always run
brain build-indexafter adding new pages or the search won't find them - Session ingest is idempotent — safe to run repeatedly
- Use
brain lintbefore committing changes to the vault - The vault is filesystem-first: you can always read/write .md files directly
- Orphan pages are INFO-level only — not errors
- Secrets in the vault are hard-blocked by the secret scanner
Verification
After setup, verify:
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckinghermes/skills/llm-wiki-brain/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/r0b0tlab/llm-wiki_obsidian_hermes_r0b0tlabbra1n/llm-wiki-brain)