Greebo Setup

Complete Claude Code configuration. Hooks, settings, memory system, token tracking, cron automation, Telegram bot, and workflow patterns.

Full setup guide: nullhex.io/guides/greebo

Quick Start

git clone https://github.com/nullhex-io/greebo-setup.git ~/greebo-setup
cd ~/greebo-setup
./sync.sh --dry-run    # Preview changes
./sync.sh              # Apply

Staying Updated

cd ~/greebo-setup
git pull && ./sync.sh

Or just check if you're behind:

./sync.sh --check

What's Included

Auto-synced (via ./sync.sh)

Category	Files	Location
Hook scripts	auto-format, extract-memories, verify-before-commit, suggest-rlm, etc.	~/.claude/hooks/
GitNexus hooks	Code intelligence integration	~/.claude/hooks/gitnexus/
Memory hooks	Session start/end, pre-message context injection	~/dev/memory/hooks/
Status line	Token tracking (raw + effective cost)	~/.claude/scripts/
Settings	Hooks config, plugins, worktree, effort level	~/.claude/settings.json (merged)

Manual setup (templates)

Category	Files	What to do
Global rules	templates/CLAUDE.md	Copy to ~/dev/CLAUDE.md, customize
Bash additions	templates/bashrc-additions.sh	Append to ~/.bashrc
tmux config	templates/tmux.conf	Copy to ~/.tmux.conf
Telegram bot	templates/claude-telegram.sh + templates/systemd/claude-telegram.service	See Telegram section below
Cron jobs	templates/cron/crontab.template	crontab -e and add entries
Thinking loop secrets	templates/thinking_loop.env.example	Copy to ~/dev/memory/thinking_loop.env, fill in
Session primer	templates/generate-primer.sh	Copy to each project's scripts/

Memory engine (copy to ~/dev/memory/engine/)

Script	Purpose	Cron
engine/thinking_loop.sh	Hourly briefing (git, health, Cortex, AI feed)	0 * * * *
engine/lint-memories.sh	Daily structural lint of memory files	0 5 * * *
engine/dream-consolidation.sh	Nightly memory synthesis and pruning	0 3 * * *

Analysis tools

Script	Purpose
scripts/token_analyzer.py	Analyze token usage across all Claude Code sessions

What's NOT Synced (personal)

• settings.local.json - your personal overrides
• settings.user.json - MCP servers with auth tokens
• Memory files - your project memories
• API keys and tokens
• Project-specific CLAUDE.md files

Setup Order

Layer 0 - Core (5 min)

1. ./sync.sh - installs hooks, settings, statusline
2. Copy templates/CLAUDE.md to ~/dev/CLAUDE.md and customize
3. Append templates/bashrc-additions.sh to ~/.bashrc
4. Copy templates/tmux.conf to ~/.tmux.conf

Layer 1 - Cortex (canonical memory) 5. Install PostgreSQL + pgvector 6. Create cortex database: sudo -u postgres createdb cortex && sudo -u postgres psql -c "CREATE EXTENSION IF NOT EXISTS vector;" cortex 7. Install Cortex plugin: /install cortex@cortex-plugins in a Claude session 8. Patch DATABASE_URL in the plugin config

Layer 2 - GitNexus (code intelligence) 9. npm install -g gitnexus && gitnexus setup 10. Index repos: cd ~/dev/myapp && gitnexus analyze

Layer 3 - Memory Freshness 11. Clone memory-freshness to ~/dev/memory-freshness, run bun install 12. sudo ln -s $(pwd)/mf.sh /usr/local/bin/mf

Memory engine 13. Copy engine/ scripts to ~/dev/memory/engine/ 14. Copy templates/thinking_loop.env.example to ~/dev/memory/thinking_loop.env and fill in secrets 15. Add cron entries from templates/cron/crontab.template

Optional 16. Set up Telegram bot using templates in templates/systemd/ and templates/claude-telegram.sh 17. Install community skills: npx skills add sickn33/antigravity-awesome-skills --all

Session Primer Pattern

Instead of running expensive codebase exploration at the start of every session, generate a compact resume document:

cp templates/generate-primer.sh ~/dev/your-project/scripts/
chmod +x ~/dev/your-project/scripts/generate-primer.sh
cd ~/dev/your-project && ./scripts/generate-primer.sh

# Auto-regenerate on every commit
cat > .git/hooks/post-commit << 'EOF'
#!/usr/bin/env bash
REPO_ROOT=$(git rev-parse --show-toplevel)
SCRIPT="${REPO_ROOT}/scripts/generate-primer.sh"
[ -x "$SCRIPT" ] && "$SCRIPT" > /dev/null 2>&1 &
EOF
chmod +x .git/hooks/post-commit

Add to your project's CLAUDE.md:

Session start: Read .planning/SESSION_PRIMER.md first. Only run full exploration for audits.

Token Analyzer

See how much your sessions cost:

python3 scripts/token_analyzer.py                    # All time
SINCE_DAYS=7 python3 scripts/token_analyzer.py       # Last 7 days
SINCE_DATE=2026-04-01 python3 scripts/token_analyzer.py  # Since a date

Options

./sync.sh              # Sync everything
./sync.sh --dry-run    # Preview without changing
./sync.sh --check      # Just check if behind
./sync.sh --hooks-only # Only sync hook scripts