openclaw-workspace-wizard

name

openclaw-workspace-wizard

description

Full guided journey to create a complete OpenClaw agent workspace from scratch. Chains discovery, memory design, persona crafting, workspace scaffolding, and quality audit into a single orchestrated flow with checkpoints. Use when the user says "create my agent from scratch", "full agent setup", "build a complete agent", "new openclaw agent", "workspace wizard", "guided agent creation", or "walk me through creating an agent". This is the comprehensive path — for quick single-file edits use openclaw-agent-architect instead.

OpenClaw Workspace Wizard

The full guided journey for creating a production-ready OpenClaw agent workspace. Chains six phases with user checkpoints between each, so you can pause, adjust, and resume at any point.

Overview

This wizard orchestrates the complete lifecycle:

Phase 1: Discovery     → What agent do you need?
Phase 2: Memory Design → How should it remember?
Phase 3: Soul Design   → Who is this agent?
Phase 4: Workspace     → What files and structure?
Phase 5: Scaffolding   → Generate everything
Phase 6: Quality Audit → Score and improve

Each phase produces artifacts that feed the next. You can skip phases if you already have the artifacts (e.g., skip Discovery if you already know exactly what you want).

Phase 0: Welcome and Setup

Configuration

Determine workspace settings:

1. Workspace name: What should this agent be called? (working title — can change later)
2. Workspace mode: Single agent or multi-agent team?
3. Workspace location: Where to create the workspace?
   • Default: ~/.openclaw/workspace/ (single-agent)
   • Multi-agent: ~/.openclaw/agents-workspaces/<agent-id>/
   • Custom path if user prefers

Resume Detection

Check the workspace location for existing artifacts:

• Discovery reports, memory architecture docs, SOUL.md drafts
• If found, present them and offer: [R] Resume from last checkpoint, [S] Start fresh, [P] Pick a specific phase

Welcome Message

"Welcome! Let's build your OpenClaw agent workspace step by step.

Your settings:

• Agent name: {name}
• Mode: {single/multi-agent}
• Output: {workspace path}

This journey has 6 phases. I'll check in with you between each one so you can adjust course. Ready?"

Wait for user confirmation.

---

Phase 1: Discovery — What Agent Do You Need?

Goal: Understand what the agent does, who it serves, and what it needs.

1.1 Reference Templates

"Before we dive in — would you like to browse reference templates for common agent archetypes? I have templates for Software Architect, Code Reviewer, Security Engineer, Technical Writer, DevOps Automator, Sprint Prioritizer, UX Researcher, and Feedback Synthesizer.

[RT] Browse templates for inspiration, or just tell me about your agent."

If RT selected: Read and present references/agent-templates/index.md, let user browse individual templates, then return here.

1.2 Intent Discovery

Ask the user:

• What is this agent's primary job? What does it do when someone talks to it?
• Who are the people that will talk to this agent?
• Where do conversations happen? (Telegram, WhatsApp, Discord, Slack, etc.)
• Is this one agent or do you envision a team of specialized agents?

Probe deeper based on responses:

• How technical is the audience?
• Will it be public-facing, team-internal, or both?
• What should the agent absolutely NOT do? (refusal boundaries)

1.3 Use Cases

Walk through 3-5 concrete scenarios:

• "Someone sends your agent a message saying X. What should happen?"
• What tasks does it handle vs delegate vs refuse?

1.4 Requirements

Based on the conversation, determine:

• Which workspace files are needed (SOUL always, IDENTITY always, AGENTS always, others optional)
• Memory strategy recommendation (minimal / standard / comprehensive)
• Whether HEARTBEAT, BOOTSTRAP, or TOOLS files are needed
• For multi-agent: initial agent roster

Discovery Checkpoint

"Phase 1 complete! Here's what I've captured:

• Agent: {name} — {one-line purpose}
• Audience: {who}
• Channels: {where}
• Mode: {single/multi}
• Files needed: {list}
• Memory: {strategy}

[C] Continue to Phase 2 (Memory Design) [R] Revise something [M] Save progress and stop here"

---

Phase 2: Memory Design — How Should It Remember?

Goal: Design the memory architecture before building the persona.

2.1 Memory Strategy

Based on the discovery:

• Minimal: MEMORY.md only, no daily logs, manual curation. Good for simple/stateless agents.
• Standard: MEMORY.md + daily logs, weekly SCRIBE compression, 7-day active window. Good for most agents.
• Comprehensive: Full architecture with archival, promotion triggers, SCRIBE compression, 14-day active window. Good for agents managing complex ongoing work.

Discuss trade-offs and confirm the strategy.

2.2 Memory Categories

What categories should MEMORY.md have?

• User Profile (almost always)
• Projects (if managing ongoing work)
• Preferences (confirmed user preferences)
• People (contacts and relationships)
• Lessons Learned (important learnings)
• Custom categories based on agent role

2.3 Daily Log Format

If daily logs enabled:

• Decisions, Facts, Actions, Open Questions sections
• For multi-agent: shared vs per-agent memory, team-memory/ directory

Memory Checkpoint

"Phase 2 complete! Memory architecture:

• Strategy: {minimal/standard/comprehensive}
• MEMORY.md categories: {list}
• Daily logs: {yes/no, format}
• Team memory: {if applicable}

[C] Continue to Phase 3 (Soul Design) [R] Revise something [M] Save progress and stop here"

---

Phase 3: Soul Design — Who Is This Agent?

Goal: Craft a coherent, opinionated, token-efficient SOUL.md.

3.1 Identity Statement

Craft collaboratively: "You are [NAME], a [ROLE] for [CONTEXT]."

One sentence. Specific. Predictive of behavior.

3.2 Core Values (3-5)

Each value needs:

• Name: Short label
• Behavioral prediction: What the agent DOES because of this value
• What it means in practice: Concrete example

Anti-patterns to avoid: "Be helpful" (too vague), "Ensure quality" (not actionable).

3.3 Boundaries

Probe for:

• What should the agent NEVER do? (hard refusals)
• What requires confirmation before action? (destructive operations)
• What's the fallback when uncertain? (ask vs guess vs refuse)

Each boundary must be specific and actionable.

3.4 Voice

Design the communication personality:

• Tone (direct/warm/casual/formal)
• Default response length
• Format preferences (bullets vs prose, emoji usage)
• Anti-patterns (phrases to never use, e.g., "Great question!")
• Optional vibe (2-3 sentences capturing personality flavor)

3.5 Token Budget Check

SOUL.md must be under 2,000 words. Present the complete draft and word count. If over, suggest specific cuts.

Soul Checkpoint

Present the complete SOUL.md draft for review.

"Phase 3 complete! Your SOUL.md is {word_count} words.

[C] Continue to Phase 4 (Workspace Design) [R] Revise something [M] Save SOUL.md and stop here"

---

Phase 4: Workspace Design — What Files and Structure?

Goal: Design IDENTITY.md and determine the complete file manifest.

4.1 IDENTITY.md

Create collaboratively:

• name: Agent display name
• emoji: Representative emoji
• status: Agent tagline

4.2 AGENTS.md Design

Design the operating manual:

• Memory management rules (write/remember/forget — informed by Phase 2)
• Safety rules (informed by SOUL.md boundaries from Phase 3)
• Group chat behavior rules
• Agent-specific workflows
• Tool rules and constraints
• Sub-agent rules (critical: any rule needed by sub-agents MUST be here)

4.3 Optional Files

For each file identified in Phase 1:

• USER.md: User profile template
• GOALS.md: Initial goal structure
• HEARTBEAT.md: Monitoring checklist (keep empty by default)
• BOOTSTRAP.md: First-run ritual
• TOOLS.md: Local environment notes

4.4 Multi-Agent Design (if applicable)

• Agent roster with per-agent directories
• Shared vs per-agent file decisions
• Routing rules and mention triggers
• Default agent and fallback behavior

Workspace Checkpoint

"Phase 4 complete! Workspace manifest: {list all files to generate} {directory structure diagram}

[C] Continue to Phase 5 (Scaffolding) [R] Revise something [M] Save progress and stop here"

---

Phase 5: Scaffolding — Generate Everything

Goal: Create all workspace files and validate them.

5.1 Create Directory Structure

For single-agent: flat structure at workspace path. For multi-agent: per-agent directories with shared files.

5.2 Generate Files

Write each file using the design decisions from Phases 1-4:

1. SOUL.md (from Phase 3 draft)
2. IDENTITY.md (from Phase 4)
3. AGENTS.md (from Phase 4 design)
4. MEMORY.md (from Phase 2 template)
5. All optional files as designed
6. Daily log template (if applicable)
7. openclaw.json (if needed)

5.3 Validate

Run checks:

• SOUL.md under 2,000 words
• AGENTS.md under 150 lines
• MEMORY.md under 200 lines (if exists)
• All cross-references consistent
• Sub-agent critical rules live in AGENTS.md (not only in SOUL.md)
• No secrets or credentials in any workspace file

Report any issues.

Scaffolding Checkpoint

"Phase 5 complete! Generated {N} files at {workspace_path}: {file list with sizes}

[C] Continue to Phase 6 (Quality Audit) [R] Fix something before audit [M] Stop here — workspace is ready for use"

---

Phase 6: Quality Audit — Score and Improve

Goal: Run a comprehensive quality audit on the generated workspace.

Invoke the openclaw-workspace-audit skill logic on the workspace we just created. This provides an independent quality check of everything we built.

Present the scored report and fix any issues identified.

Completion

"Your OpenClaw workspace is ready!

Workspace: {path} Quality Score: {score}/200 — {rating} Files: {list}

Next steps:

1. Start a session to test your new agent
2. After a week of use, run /oc-workspace-audit to reassess quality
3. Run /oc-autonomy if you want to enable autonomous operation
4. Run /oc-memory if you want to fine-tune memory architecture

Quick commands:

• /oc-architect — Modify individual workspace files
• /oc-workspace-audit — Re-run quality audit anytime
• /oc-autonomy — Autonomy readiness assessment"

Integration

This wizard uses content and patterns from:

• openclaw-agent-architect — File generation patterns and interview techniques
• openclaw-agent-builder — Guardrails checklist and autonomy design
• openclaw-memory — Four-layer memory model and SCRIBE compression
• openclaw-workspace-audit — Scored quality assessment
• Reference agent templates — Pre-built personas for inspiration (in references/agent-templates/)