The Lab
Agent roster, routing logic, automation maturity, and ownership gaps across the system.
What this section is now
The Lab is the system map for who should handle what. Its job is to show the real agent lineup, where work should route, how mature the automations are, and where ownership is missing. It should help decide the right owner fast, not just list agents decoratively.
Core agents
Configured in OpenClaw
5
Open agent tasks
Non-done tasks assigned to agents
1
Unowned projects
Active work with no agent owner yet
2
Active automation runs
Queued, running, or waiting approval
2
Who should handle what
Fast routing guide so new work lands with the right agent.
Claw (Main Agent)
DefaultChief of staff, orchestrator, entry point for everything
Use for coordination, decisions, cross-section work, and anything that needs orchestration.
Scout (Research Agent)
Market research and product analysis
Use for market research, competitor digging, product validation, and source gathering.
Forge (Coding Agent)
Development and coding tasks
Use for product builds, code changes, debugging, and implementation work.
Sentinel (Mac Agent)
Mac system management and monitoring
Use for machine health, maintenance, local automation checks, and system issues.
Coding Agent
Dedicated coding workspace
Use for heavier isolated coding work when the main workspace should stay clean.
Ownership gaps
These are the places where routing is still weak or waiting.
Priority projects without owner
Get Things Done Software
idea · needs agent owner
Tasks waiting review
Review the new Tasks section
Mustafa · waiting review
Routing note
If this list grows, The Lab should not absorb the work itself — it should make the missing owner obvious, then hand the work off to Tasks, Projects, or Ecom.
Core agent roster
The actual OpenClaw team configured on this machine, with ownership and load visibility.
Claw (Main Agent)
Defaultgpt-5.4Light loadChief of staff, orchestrator, entry point for everything
General conversation, business decisions, task delegation
Open tasks
1
Projects
0
Workspace
~/.openclaw/workspace
Scout (Research Agent)
gpt-5.4Light loadMarket research and product analysis
Product research, competitor analysis, market trends
Open tasks
0
Projects
0
Workspace
~/.openclaw/workspace/agents/research
Forge (Coding Agent)
gpt-5.4Light loadDevelopment and coding tasks
Code development, debugging, technical implementation
Open tasks
0
Projects
0
Workspace
~/.openclaw/workspace/agents/forge
Sentinel (Mac Agent)
glm-4.5-air:freeLight loadMac system management and monitoring
System monitoring, maintenance, alerts
Open tasks
0
Projects
0
Workspace
~/.openclaw/workspace/agents/mac
Coding Agent
gpt-5.4Light loadDedicated coding workspace
Complex coding projects, separate environment
Open tasks
0
Projects
0
Workspace
~/openclaw/workspace-coding
Ecom automation lane
Specialized automations that belong inside Ecom OS, not the core assistant roster.
Research Agent
Competitor, market, offer, and hook research
Total runs
1
Active now
0
Scope
Best first integrated agent for Ecom OS.
Creative Agent
Hooks, scripts, ad concepts, and creative briefs
Total runs
1
Active now
1
Scope
Active run(s)
Hook drafting batch
Generating VSL-friendly hooks from research findings.
Should attach outputs directly to products.
Facebook Ads Agent
Read account state, suggest actions, and later control campaigns safely
Total runs
0
Active now
0
Scope
Phase 1 should be read/report only.
Ops Agent
Logs work, saves outputs, and keeps the system state clean
Total runs
1
Active now
1
Scope
Active run(s)
Launch prep playbook
Prepared the final pre-launch checklist and next actions.
This is the glue layer for the Ecom OS.
Ready Ecom agents: 1 of 4. The rest are intentionally shown as planned or needing integration so this page stays honest.
What belongs here
Use The Lab when you need the system map.
- Which agents exist
- Which agent should own a new kind of work
- How mature the automation is
- Where ownership is missing or overloaded
What does not belong here
Don’t turn The Lab into a catch-all.
- Task execution history → Tasks / Progress
- Product-specific work → Ecom OS
- Machine health → Dashboard / Metrics
- Loose notes or scratch thinking → Notes
Best next upgrades
Useful later, not required for this v1.
- Live OpenClaw session/status integration
- Direct launch/run actions for agents
- TaskFlow run visibility across sections
- Smart owner suggestions when new projects/tasks are created