Claude Code CLI - Infrastructure Configuration

Directory Structure

~/.claude/                              # USER LEVEL (personal preferences)
├── config.toml                         # Trust levels, model preferences
├── settings.json                       # Always-on features, global permissions
├── commands/                           # User slash commands (6 commands)
│   ├── cmemory.md                      # /cmemory - Save to OpenMemory
│   ├── cdebug.md                       # /cdebug - Enable debug mode
│   └── ...
├── skills/                             # User skills (7 skills)
│   ├── architecture-diagram-creator/
│   ├── gitlab-integration/
│   ├── naming-validator/
│   ├── secrets-security/
│   ├── skill-creator/
│   └── ...
├── agents/                             # User agents (1 agent)
│   └── pm.md                           # Project Manager agent
└── session-env/                        # Per-session environment snapshots

~/projects/.claude/                     # PROJECT LEVEL (infrastructure-wide)
├── settings.local.json                 # 174 pre-approved permissions
├── init.md                             # Bootstrap file (loads memoryfiles.md)
├── commands/                           # Project slash commands (4 commands)
│   ├── cexit.md                        # /cexit - Session exit summary
│   ├── cproject-health.md              # /cproject-health
│   └── ...
├── skills/                             # Infrastructure skills (8 skills)
│   ├── keycloak-setup/
│   ├── traefik-setup/
│   ├── postgres-integration/
│   ├── service-deployment/
│   ├── troubleshooting/
│   └── ...
├── agents/                             # Infrastructure agents (4 agents)
│   ├── architect.md                    # System architecture specialist
│   ├── developer.md                    # Implementation specialist
│   ├── security.md                     # Security & network specialist
│   └── qa.md                           # Quality assurance specialist
└── docs/                               # Agent documentation

Configuration Files

config.toml

~/.claude/config.toml

User-level preferences and trust settings

Setting	Value
model	gpt-5-codex
/home/administrator	trusted
/home/administrator/projects	trusted

settings.json

~/.claude/settings.json

Global settings and features

Setting	Value
alwaysThinkingEnabled	true
enableAllProjectMcpServers	true
permissions.deny	Read(/projects/administrator/projects/claude)

settings.local.json - Pre-Approved Permissions

~/projects/.claude/settings.local.json

174 pre-approved tool permissions for autonomous operation

85+

Bash Commands

7

Read Paths

8

WebFetch Domains

Categories: Docker operations, Git workflows, File operations, Network diagnostics, Database access, System administration, MCP tools, Skills, Slash commands

MCP Integration

code-executor - Primary MCP Server

docker exec -i mcp-code-executor npx tsx /app/mcp-server.ts

Claude Code
CLI

→

code-executor
MCP Server

→

MCP Proxy
:9090

→

9 Backend Servers
71 Tools

Server	Tools	Purpose
postgres	9	Database queries, schema info, table stats
filesystem	9	File operations within /workspace
arangodb	7	Document/graph database for AI memory
memory	9	Knowledge graph storage
minio	9	S3-compatible object storage
ib	10	Interactive Brokers market data
playwright	6	Browser automation, screenshots
n8n	6	Workflow automation
timescaledb	6	Time-series database

Skills (15 Total)

User Skills (7)

~/.claude/skills/

🛠
architecture-diagram-creator
Create project architecture diagrams
🔗
gitlab-integration
GitLab repo management, MRs, CI/CD
✅
naming-validator
Validate resource naming standards
🔒
secrets-security
Secrets management, security hardening
📚
skill-creator
Guide for creating new skills
📝
claudemd-generator
Generate CLAUDE.md files
🧠
openmemory
OpenMemory (mem0) integration

Infrastructure Skills (8)

~/projects/.claude/skills/

🔐
keycloak-setup
OIDC client configuration for SSO
📐
traefik-setup
Reverse proxy routing, SSL/TLS
🗃
postgres-integration
Database setup, user management
🚀
service-deployment
Docker deployment orchestration
🔧
troubleshooting
Diagnostic procedures, log analysis
📊
logging-observability
Loki, Grafana, monitoring config
🌐
network-config
Docker networks, DNS, topology
📄
nginx-static-site
Static site deployment guide

Specialized Agents (4 Total)

User Agent

~/.claude/agents/

💼
pm (Project Manager)
Multi-step project orchestration, coordinates architect/security/developer agents, tracks progress

Infrastructure Agents

~/projects/.claude/agents/

🛠
architect
System design, technology stack decisions, integration planning, performance optimization
💻
developer
Implementation, deployment, debugging, database operations, container management
🛡
security
Network topology, firewall, SSL/TLS, Keycloak/OAuth2, Traefik routing, security audits

Slash Commands (10 Total)

User Commands (6)

~/.claude/commands/

💾
/cmemory
Save facts and lessons to OpenMemory (mem0)
🔎
/cdebug
Enable debug mode for troubleshooting
🚫
/cundebug
Disable debug mode
✅
/cvalidate-claude
Validate Claude configuration

Project Commands (4)

~/projects/.claude/commands/

🚪
/cexit
Save session summary before exiting
💙
/cproject-health
Check project health status
🔄
/cprojectswitch
Switch between projects
🚀
/cvalidate-deploy
Validate deployment readiness

Context Loading Flow

SESSION START
     |
     v
+---------------------------+
|  Layer 0: AI Model        |  Claude Opus 4.5 base knowledge
+---------------------------+
     |
     v
+---------------------------+
|  Layer 1: User Context    |  ~/.claude/config.toml
|                           |  ~/.claude/settings.json
|                           |  User commands, skills, agents
+---------------------------+
     |
     v
+---------------------------+
|  Layer 2: Project Context |  ~/projects/.claude/init.md
|                           |  → @memoryfiles.md (9 AINotes files)
|                           |  → CLAUDE.md per-project files
|                           |  Project commands, skills, agents
+---------------------------+
     |
     v
+---------------------------+
|  Layer 3: Permissions     |  settings.local.json
|                           |  174 pre-approved operations
|                           |  Autonomous operation enabled
+---------------------------+
     |
     v
+---------------------------+
|  Layer 4: Runtime Tools   |  MCP code-executor
|                           |  71 tools across 9 servers
|                           |  Skills loaded on-demand
+---------------------------+

Key Paths Reference

Purpose	Path
User config	`~/.claude/config.toml`
User settings	`~/.claude/settings.json`
Project permissions	`~/projects/.claude/settings.local.json`
Bootstrap file	`~/projects/.claude/init.md`
Session history	`~/.claude/history.jsonl`
MCP config	`claude mcp list`
Project CLAUDE.md	`~/projects/CLAUDE.md`
AINotes (context)	`~/projects/AINotes/*.md`