Team Wiki/setup

Team-Wiki Setup

Initializes the Team-Wiki shared knowledge directory structure and core files. Creates the domain-specific schema, index, log, and folder scaffolding required for GBrain sync.

Prerequisites

• GBrain installed and initialized (gbrain init already run, database at ~/.hermes/shared-brain/brain.pglite)
• Obsidian vault directory exists
• WIKI_PATH environment variable set (e.g., ~/Hermes Vault/Hermes/Team-Wiki)
• User has confirmed Team-Wiki is separate from personal PARA vault

What This Does

Creates the following structure under WIKI_PATH:

Team-Wiki/
├── SCHEMA.md          # Domain conventions, tag taxonomy, page types
├── index.md           # Alphabetical catalog of all entities
├── log.md             # Append-only chronological action log
├── entities/
│   ├── README.md
│   ├── agents/        (foreman, coder, architect, debugger, reviewer, secretary, researcher, devops, data-analyst, security, senna)
│   ├── people/        (andrej-karpathy, garry-tan, …)
│   ├── companies/     (anthropic, openai, 37signals, …)
│   └── projects/      (hermes-agent, team-wiki, hermes-workspace, …)
├── concepts/          (pglite, mitre-attack, rag, …)
├── comparisons/       (hermes-vs-swarmclaw, …)
└── queries/           (research questions & answers)

Writes domain-specific conventions into SCHEMA.md (Hermes ecosystem: agents, skills, infrastructure, companies, people, projects). Sets up tag taxonomy using domain:item prefixes.

Invocation

# With WIKI_PATH set
skill:team-wiki/setup

# Or explicit path
skill:team-wiki/setup --wiki-path ~/Hermes\ Vault/Hermes/Team-Wiki

Post-Setup

1. Review SCHEMA.md and adjust taxonomy tags to match your domain
2. Populate initial index.md with any existing known entities
3. Ensure GBrain MCP server is configured in ~/.hermes/config.yaml
4. Run gbrain sync --follow --repo $WIKI_PATH to start continuous sync (or configure Hermes cron)

Verification

• All expected directories exist under WIKI_PATH
• SCHEMA.md contains domain: Hermes agent ecosystem and tag taxonomy
• log.md contains initialization entry with timestamp
• index.md has placeholder sections matching SCHEMA categories

Notes

• This skill is idempotent: running it twice is safe (skips existing files)
• Do NOT migrate existing PARA vault notes into Team-Wiki unless explicitly requested; they remain separate
• The schema follows llm-wiki patterns: MECE directories, compiled truth + timeline layers, enrichment on every signal