openclawclaude-codev1.0.0

Claude Family

@TalkingMonkeyOz1 stars· last commit 1mo ago· 1 open issues

Persistent identity and memory system for coordinating multiple Claude instances with PostgreSQL backend and MCP memory sync

7.1/10
Verified
Mar 9, 2026

// RATINGS

GitHub Stars

New / niche

🟢ProSkills ScoreAI Verified
7.1/10
📍

Not yet listed on ClawHub or SkillsMP

// README

# Claude Family - Infrastructure Project **Type**: Infrastructure **Purpose**: Shared configuration, scripts, and documentation for Claude instances --- ## What Is This? The **Claude Family** project provides shared infrastructure for multiple Claude instances working across different projects with persistent knowledge storage in PostgreSQL. **Key Features:** - PostgreSQL database (`ai_company_foundation`) for persistent knowledge - MCP servers (postgres, memory, filesystem, tree-sitter, github) - Shared documentation and scripts - Session history tracking - Knowledge management system --- ## Current Architecture (Oct 21, 2025) **Two Active Identities:** 1. **claude-desktop** - GUI interface (Claude Desktop app) 2. **claude-code-unified** - CLI interface (Claude Code Console) **Project-Aware Context:** - Each project has its own `CLAUDE.md` file that auto-loads - Shared scripts and documentation in `claude-family` repository - Universal knowledge stored in PostgreSQL - MCP servers available to all instances --- ## Directory Structure ``` claude-family/ ├── docs/ # Documentation and session notes ├── .claude/ # Shared slash commands │ └── commands/ │ ├── session-start.md # Startup checklist │ └── session-end.md # End-of-session logging ├── scripts/ │ ├── audit_docs.py # Documentation audit │ ├── install_git_hooks.py # Install git hooks │ ├── pre-commit-hook.sh # Enforce CLAUDE.md limits │ ├── auto_sync_startup.py # PostgreSQL → MCP sync │ ├── load_claude_startup_context.py │ └── sync_postgres_to_mcp.py ├── .docs-manifest.json # Documentation registry ├── CLAUDE.md # Project context (auto-loaded) └── README.md # You are here ``` --- ## Quick Start ### Every Session (Optional) Run `/session-start` to: - Check documentation health (monthly) - Query for existing solutions before proposing new ones - Search memory graph for relevant context ### End of Session (Optional) Run `/session-end` to: - Log session to PostgreSQL - Store reusable knowledge - Update memory graph --- ## PostgreSQL Database **Database**: `ai_company_foundation` **Schemas:** - `claude_family` - Identities, session history, shared knowledge - `nimbus_context` - Nimbus project context - `public` - Work packages, projects, SOPs **Key Tables:** - `claude_family.identities` - Claude instances - `claude_family.session_history` - Session logs - `claude_family.shared_knowledge` - Reusable patterns/learnings - `claude_family.project_workspaces` - Project locations --- ## MCP Servers (Always Available) - **postgres** - Database access, session logging - **memory** - Persistent memory graph - **filesystem** - File operations - **tree-sitter** - Code structure analysis - **github** - GitHub operations - **sequential-thinking** - Complex problem solving --- ## Documentation Management **System**: Simple 3-component system 1. **`.docs-manifest.json`** - Single source of truth for all markdown files 2. **`scripts/audit_docs.py`** - Monthly audit script 3. **Git pre-commit hook** - Enforces CLAUDE.md ≤250 lines **Install hook:** ```bash python scripts/install_git_hooks.py ``` **Run audit:** ```bash python scripts/audit_docs.py ``` **Rules:** - CLAUDE.md must stay ≤250 lines (enforced automatically) - Deprecated docs kept for 90 days, then archived - Audit monthly to prevent bloat --- ## Benefits | Before | After | |--------|-------| | 30-60 min context reload | Auto-loads from CLAUDE.md | | No memory across reboots | Permanent PostgreSQL storage | | Repeated explanations | Persistent knowledge & patterns | | Mixed work/personal projects | Clean schema separation | --- ## Troubleshooting **Problem:** "Graph is empty" after reboot **Solution:** Use MCP postgres tool to query shared_knowledge directly **Problem:** Python script errors **Solution:** Check PostgreSQL is running **Problem:** Can't commit CLAUDE.md changes **Solution:** CLAUDE.md exceeds 250 lines - move content to docs/ subdirectory --- ## Architecture History - **Oct 10, 2025**: Initial 9-identity architecture with isolated workspaces - **Oct 21, 2025**: Simplified to 2 identities with project-aware context - **Oct 23, 2025**: Added documentation management system See `docs/DEPRECATED_OCT10_ARCHITECTURE.md` for historical architecture. --- ## Future Enhancements - Automatic session history recording - Semantic search with vector embeddings - ClaudePM integration for centralized documentation - PostgreSQL automated backups - Cross-Claude task handoff workflows --- **Created:** 2025-10-10 **Last Updated:** 2025-10-23 **Location:** `C:\Projects\claude-family\`

// REPO STATS

1 stars
1 open issues
Last commit: 1mo ago

// PROSKILLS SCORE

7.1/10

Good

BREAKDOWN

Code Quality6.5/10
Documentation7.5/10
Functionality7/10
Maintenance7.5/10
Security6/10
Uniqueness8/10
Usefulness7.5/10

// DETAILS

Categorycoding
Versionv1.0.0
PriceFree