Openclaw Architect

Design, configure, debug, and optimize OpenClaw AI agent deployments. Master guide for gateway configuration, openclaw.json settings, model routing and fallb...

Security Scan

The skill's files, instructions, and required environment variable line up with its stated purpose of auditing and health-checking an OpenClaw deployment; nothing obvious is requesting unrelated credentials or installing remote code.

Purpose & Capability

Name/description (OpenClaw config analysis, health checks, docs) match the included SKILL.md, reference docs, and the two Python scripts. The single required env var (OPENCLAW_WORKSPACE) is appropriate for locating the workspace.

Instruction Scope

SKILL.md instructs the agent to run the included scripts which read local config and workspace files. The scripts perform local health checks (HTTP calls to localhost ports, file existence checks, pgrep, docker/tailscale checks) and examine openclaw.json. They do not contact remote third-party endpoints. Note: health_check runs shell commands (via subprocess.run with shell=True) and accesses system paths (e.g., /var/run/docker.sock, /proc), and outputs diagnostics to stdout; running these will reveal local config/state to whoever receives the script output.

Install Mechanism

No install spec — instruction-only with included stdlib Python scripts. Nothing is downloaded or written to disk by an installer step.

Credentials

Only OPENCLAW_WORKSPACE is required (declared as primaryEnv), which is proportional. However the health checker probes system services (Docker socket, Qdrant, Neo4j, local gateway ports) and runs system commands; it may therefore require filesystem and process visibility and could reveal sensitive local information in its output. No unrelated external API keys or unrelated env variables are requested.

Persistence & Privilege

always is false and the skill does not request permanent/system-wide modifications. It does not modify other skills or agent configs; it only reads local files and executes diagnostic scripts when invoked.

Assessment

This skill appears coherent for auditing and health-checking an OpenClaw deployment. Before installing or running it: 1) Verify OPENCLAW_WORKSPACE points to the intended workspace. 2) Inspect the two included scripts (config_analyzer.py and health_check.py) yourself — they are stdlib Python and readable. 3) Run health checks locally and in a trusted environment because the health_check script executes shell commands and checks files/sockets (Docker, /var/run/docker.sock, tailscale, local HTTP endpoints) and will print local configuration and status to stdout. 4) The analyzer will warn about API keys found in openclaw.json (it reports key locations, not values); if you keep secrets in config files, consider moving them to environment variables or secret storage. 5) Only proceed if you trust the skill source and are comfortable exposing the resulting diagnostic output to the agent or logs where the agent runs.

The definitive skill for understanding, configuring, debugging, and optimizing OpenClaw deployments. Built from real production experience.

Features

• Analyze configurations — audit openclaw.json and suggest improvements
• Health check systems — validate all OpenClaw subsystems in one command
• Configure model routing — set up primary models, fallback chains, cost tiers
• Build skills — SKILL.md format, CLI design, publishing to ClawHub
• Debug gateway issues — troubleshoot errors, cron failures, session crashes
• Optimize performance — model selection, cost reduction, context management
• Manage cron jobs — scheduling, error handling, retry patterns
• Configure memory systems — Qdrant, Neo4j, SQLite integration
• Deploy infrastructure — Docker, Tailscale VPN, networking
• Post-update verification — checklist for safe OpenClaw upgrades

Requirements

Variable	Required	Description
OPENCLAW_WORKSPACE	✅	Path to OpenClaw workspace directory

Quick Start

PY=~/.openclaw/workspace/.venv/bin/python3

# Analyze your openclaw.json configuration
$PY skills/openclaw-architect/scripts/config_analyzer.py

# Health check all OpenClaw systems
$PY skills/openclaw-architect/scripts/health_check.py

Commands

Config Analyzer

# Audit current configuration
$PY skills/openclaw-architect/scripts/config_analyzer.py

# Analyze a specific config file
$PY skills/openclaw-architect/scripts/config_analyzer.py --config /path/to/openclaw.json

Health Check

# Check all subsystems
$PY skills/openclaw-architect/scripts/health_check.py

# Check specific subsystem
$PY skills/openclaw-architect/scripts/health_check.py --check gateway
$PY skills/openclaw-architect/scripts/health_check.py --check cron
$PY skills/openclaw-architect/scripts/health_check.py --check memory

References

File	Description
references/architecture-overview.md	How OpenClaw works end-to-end
references/config-reference.md	All openclaw.json options documented
references/skills-guide.md	Building and publishing skills
references/cron-guide.md	Cron job scheduling and patterns
references/memory-guide.md	Memory system configuration
references/troubleshooting.md	Common fixes and debugging
references/optimization-tips.md	Performance tuning guide

Architecture Principles

1. Brain-First — Strategic content → Mem0/Qdrant/Neo4j/SQLite. Markdown = operational logs only.
2. Fault-Tolerant — Always configure 2+ fallback models. Test each one works.
3. Credit-Aware — Monitor usage, auto-switch tiers, alert before exhaustion.
4. Skills = Publishing — Every reusable pattern becomes a ClawHub skill.
5. Self-Monitoring — System watches its own health, uptime, costs via cron.
6. Automate Repetition — If it happens twice → cron job or script.

Script Reference

Script	Description
{baseDir}/scripts/config_analyzer.py	Audits openclaw.json configuration
{baseDir}/scripts/health_check.py	Validates all OpenClaw subsystems

Output Format

All commands output structured text with clear pass/fail indicators and actionable recommendations.

Data Policy

This skill reads local configuration files only. No data is sent to external services.

---

Built by M. Abidi

LinkedIn · YouTube · GitHub · Book a Call