OperatorOne

Built by Dennis and Bennett.

OperatorOne is a multi-agent OpenClaw system for running a full startup execution loop:

idea discovery → landing + demand generation → outreach + conversion → feedback-driven iteration

Current north-star KPI: first $100 MRR.

---

Agent topology

OperatorOne has 4 specialist execution agents plus a manager control-plane workspace.

Workspace	Role	In openclaw/agents.manifest.json	Primary responsibility
workspaces/op1_product	Product	✅	Idea discovery, web product build/deploy, landing handoff
workspaces/op1_marketing	Marketing	✅	SEO/content/campaign pipeline
workspaces/op1_sales	Sales	✅	Prospecting, outreach, conversion
workspaces/op1_operations	Operations	✅	KPI tracking, feedback processing, iteration loop
workspaces/op1_manager	Manager (control plane)	❌	Cross-agent orchestration, audits, documentation hygiene

Sync scripts register the 4 specialist agents by default. op1_manager is intentionally kept as a local control-plane workspace.

---

End-to-end flow

Primary chain:

1. op1_product writes handoffs/product_to_marketing.json
2. op1_marketing writes handoffs/marketing_to_sales.json
3. op1_sales writes handoffs/sales_to_operations.json
4. op1_operations writes:
   • handoffs/operations_to_product.json
   • handoffs/operations_to_marketing.json
   • handoffs/operations_to_sales.json

Iteration loop (Stage3 Ops):

• handoffs/operations_to_product_iterate.json
• handoffs/operations_to_marketing_iterate.json
• handoffs/operations_to_sales_iterate.json

See handoffs/README.md for contract details.

---

Repository layout

• openclaw/ — profile sync + safety scripts (operatorone profile)
• workspaces/ — isolated agent workspaces and stage artifacts
• handoffs/ — inter-agent JSON contracts
• dashboard/ — local monitor + venture studio web app
• official-site/ — OperatorOne official project introduction page (landing/login-style entry page)
• shared/ — shared prompts/skills/templates
• docs/ — architecture, collaboration protocol, runbook

---

Quick start (local, profile isolation)

1. Clone into:
   • ~/.openclaw/workspace/OperatorOne
2. Run safe sync:
   • bash openclaw/sync-operatorone-safe.sh
3. Verify profile and agents:
   • openclaw --profile operatorone status
   • openclaw --profile operatorone agents list
4. Start a fresh chat session (/new)

Advanced:

• Config-only sync (no service action): bash openclaw/sync-openclaw.sh
• Temporary dual-gateway mode: bash openclaw/sync-operatorone-safe.sh --allow-dual-gateway

---

Z.AI (GLM) core integration

OperatorOne is configured to use Z.AI GLM as the default core model path:

• Primary: zai/glm-4.6
• Fallback: openai-codex/gpt-5.3-codex (safety fallback only)

Secure setup (local only, never commit API keys)

export ZAI_API_KEY='YOUR_ZAI_KEY'
bash scripts/setup_zai_core.sh --profile operatorone --probe-all-agents

What this does:

1. Stores ZAI_API_KEY in the local operatorone profile config (outside repo runtime)
2. Sets default model to zai/glm-4.6
3. Syncs fallback list from openclaw/agents.manifest.json
4. Generates an integration verification report:
   • docs/evidence/zai/core_integration_report.latest.json

Manual verification (without probes):

python3 scripts/verify_zai_core_integration.py --repo-root . --profile operatorone

Strict runtime probe (temporarily disables fallbacks, then restores):

python3 scripts/verify_zai_core_integration.py \
  --repo-root . \
  --profile operatorone \
  --probe-all-agents \
  --probe-without-fallback

Troubleshooting:

• If you don’t see successful Z.AI completions in the Z.AI console, check probe reports for plan/limit errors (for example: 429 ... does not yet include access to GLM-5).
• In that case, keep zai/glm-4.6 as primary (or switch to another available GLM tier in your plan).
• The system will fall back to the safety model unless fallbacks are temporarily disabled for strict verification.

---

Dashboard

Run local dashboard:

python3 dashboard/server.py --host 127.0.0.1 --port 8765

Open:

• http://127.0.0.1:8765

Details:

• dashboard/README.md
• docs/studio_usage_guide.md

---

Official project introduction page

OperatorOne includes an official intro page in official-site/.

Purpose:

• Introduce the project clearly to collaborators, prospects, and new team members.
• Explain multi-agent capabilities, handoff loop, and value proposition on one page.

Local preview:

cd official-site
python3 -m http.server 4173

Open:

• http://127.0.0.1:4173

Current live URL:

• https://official-site-theta.vercel.app

---

Workspace docs

• Product: workspaces/op1_product/README.md
• Marketing: workspaces/op1_marketing/README.md
• Sales: workspaces/op1_sales/README.md
• Operations: workspaces/op1_operations/README.md
• Manager: workspaces/op1_manager/README.md

---

Project docs map

• Architecture: docs/architecture.md
• Collaboration protocol: docs/collaboration.md
• Ops runbook: docs/runbook.md
• Full doc index: docs/README.md

Contract validation helpers:

• python3 scripts/validate_handoffs.py --repo-root .
• python3 scripts/upgrade_handoffs.py --repo-root .

Change hygiene helpers:

• python3 scripts/change_hygiene_guard.py --staged (check staged set for mixed source+generated churn)
• python3 scripts/reset_generated_artifacts.py (dry-run noisy generated deltas)
• python3 scripts/reset_generated_artifacts.py --apply (restore noisy generated deltas)

---

Note on generated artifacts

workspaces/*/research/** contains many *.latest.* and run snapshots that are intentionally machine-updated. When reviewing diffs, separate code/docs/contract changes from pipeline output refreshes.