MH1 — Your CMO Co-Pilot

Version 0.6.0 | Status: Production | Last Updated: January 30, 2026

MH1 is an AI-powered marketing operations CLI. It acts as your CMO co-pilot — executing skills, managing clients, and delivering results without the technical complexity.

███╗   ███╗██╗  ██╗  ██╗
████╗ ████║██║  ██║  ███║
██╔████╔██║███████║  ╚██║
██║╚██╔╝██║██╔══██║   ██║
██║ ╚═╝ ██║██║  ██║   ██║
╚═╝     ╚═╝╚═╝  ╚═╝   ╚═╝

---

Install

One-liner (Recommended)

curl -fsSL https://raw.githubusercontent.com/NewGameJay/mh1/main/install.sh | bash

This installs to ~/.mh1 and adds mh1 to your PATH. Then just:

mh1

pip Install (Alternative)

pip install mh1-copilot
mh1

On first run, it downloads the full system to ~/.mh1.

Manual Install

git clone https://github.com/NewGameJay/mh1.git ~/.mh1
cd ~/.mh1
./install.sh

First Run

On first run, MH1 will ask for your Anthropic API key:

ANTHROPIC_API_KEY not found.
Enter your API key: sk-ant-...

Or set it beforehand:

echo "ANTHROPIC_API_KEY=sk-ant-your-key" >> ~/.mh1/.env

That's it. Run mh1 and it will guide you from there.

---

What Can MH1 Do?

70+ Marketing Skills

Category	Examples
Analysis	Lifecycle audit, churn prediction, cohort retention, pipeline analysis
Research	Competitor research, company research, founder research
Content	Email copy, newsletters, ghostwriting, SEO content
Strategy	Positioning angles, email sequences, GTM engineering
Operations	Client onboarding, data quality audit, account 360

Intelligent Workflows

MH1 automatically detects what you need:

You Say	MH1 Does
"Run a lifecycle audit"	Executes single skill
"Acme Corp"	Starts client onboarding
"Comprehensive retention strategy"	Creates module with multiple skills
"Connect HubSpot"	Guides platform configuration
"What skills help with churn?"	Answers directly

---

How It Works

Client-Centric

Everything is organized by client:

clients/
└── acme-corp/
    ├── config/        # Platform connections, thresholds
    ├── data/          # Raw data files
    └── reports/       # Skill outputs (markdown)

Skill Execution

1. You request → "run lifecycle audit"
2. MH1 checks → Do we have required inputs? Platform access?
3. MH1 asks → Collects anything missing
4. You confirm → Review inputs before execution
5. MH1 executes → Streams progress in real-time
6. Output saved → clients/{name}/reports/lifecycle-audit-{timestamp}.md

Module Workflow (Complex Tasks)

For tasks needing 3+ skills:

1. Module created → modules/{name}-{date}/
2. MRD generated → Requirements document
3. Plan created → Skill sequence with dependencies
4. You approve → Review before execution
5. Skills execute → Sequential with checkpoints
6. Outputs compiled → All deliverables in one place

---

CLI Commands

Command	Description
./mh1	Start interactive mode
./mh1 --client "Name"	Start with a client
1	List all skills
2	List all agents
3	Switch or create client
?	Show help
q	Quit

Example Sessions

Single Skill:

> run lifecycle audit

To run lifecycle-audit, I need:
- HubSpot access for contact data
- (Optional) Snowflake for usage enrichment

Do you have HubSpot connected? What contact limit?

> Yes, limit to 100 contacts

Running lifecycle-audit with:
- Platform: HubSpot (connected)
- Limit: 100 contacts

Proceed? [Y/n]

New Client:

> Acme Corp

Detected new client: Acme Corp

Starting onboarding...

Company name: Acme Corp
Industry: SaaS
Website: https://acme.com
Your role: Head of Marketing

What CRM do you use?
1. HubSpot
2. Salesforce
...

---

Configuration

Environment Variables (.env)

# Required
ANTHROPIC_API_KEY=sk-ant-...

# Optional - Platform connections
HUBSPOT_API_KEY=...
SNOWFLAKE_ACCOUNT=...
SNOWFLAKE_USER=...
SNOWFLAKE_PASSWORD=...

Client Config (clients/{name}/config/)

Platform connections and thresholds are stored per-client:

# datasources.yaml
warehouse:
  type: snowflake
  database: "CLIENT_DB"
crm:
  type: hubspot
thresholds:
  high_value_min: 10000
  dormant_days: 30

---

Project Structure

mh1-hq/
├── mh1                     # CLI entry point (run this)
├── install.sh              # One-liner installer
├── .env                    # Your API keys (auto-created on first run)
├── requirements.txt        # Python dependencies
│
├── clients/                # Client data & outputs
│   └── {client-name}/
│       ├── config/         # Platform configs
│       ├── data/           # Raw data
│       └── reports/        # Skill outputs (markdown)
│
├── modules/                # Complex project folders
│   └── {module-name}/
│       ├── MRD.md          # Requirements document
│       ├── .plan.md        # Execution plan
│       ├── state.json      # Progress tracking
│       └── outputs/        # Deliverables
│
├── .skills/                # 70+ marketing skills
│   ├── analysis-skills/
│   ├── generation-skills/
│   ├── strategy-skills/
│   └── ...
│
├── agents/                 # AI agent definitions
│   ├── orchestrators/
│   ├── workers/
│   └── evaluators/
│
├── lib/                    # Core library
│   ├── workflow/           # CLI workflow components
│   ├── evaluator.py        # Quality evaluation
│   ├── budget.py           # Cost tracking
│   └── ...
│
├── config/                 # System configuration
│   ├── model-routing.yaml
│   ├── input_schemas.yaml
│   └── quotas.yaml
│
└── prompts/                # System prompts
    └── system/
        └── mh1-cmo-copilot.md

---

For Developers

Adding Skills

Skills live in .skills/{category}-skills/{skill-name}/:

.skills/analysis-skills/my-skill/
├── SKILL.md          # Definition (YAML frontmatter + markdown)
├── schemas/
│   ├── input.json    # Input validation
│   └── output.json   # Output validation
└── examples/

Skill Definition (SKILL.md)

---
name: my-skill
description: What this skill does
inputs:
  - name: param1
    type: string
    required: true
outputs:
  result: object
---

# Skill: My Skill

## When to Use
...

## Process
1. Step one
2. Step two
...

Workflow Components (lib/workflow/)

Module	Purpose
pathway.py	Detects which workflow to use
inputs.py	Structured input collection
markers.py	Parses [[SKILL:name]] markers
module_manager.py	Creates/manages module folders
skill_executor.py	Executes skills via Claude

---

Quality & Cost

Quality Gates

Every skill output is evaluated:

• Schema validation
• Factuality check
• Completeness check
• Brand voice match

Score ≥ 80% → Auto-deliver Score < 80% → Refinement or review

Cost Optimization

MH1 uses intelligent model routing:

Task	Model	Cost
Extraction	Claude Haiku	$
Analysis	Claude Haiku	$
Synthesis	Claude Sonnet	$$$
Strategy	Claude Sonnet	$$$

Result: 60-70% savings vs using premium models for everything.

Budget Tracking

Per-client cost limits in config/quotas.yaml:

defaults:
  daily_limit_usd: 100
  monthly_limit_usd: 2000

---

Troubleshooting

"ANTHROPIC_API_KEY not set"

Make sure .env exists and contains your key:

cat .env
# Should show: ANTHROPIC_API_KEY=sk-ant-...

"Skill not found"

Check available skills:

./mh1
# Then press 1 to list skills

"Missing required inputs"

MH1 now validates inputs before execution. Provide the required inputs when asked.

"MCP connection failed"

Check platform credentials in .env or clients/{name}/config/.

---

Changelog

v0.6.0 (January 30, 2026)

CMO Co-Pilot CLI:

• New interactive CLI with CMO co-pilot persona
• Client-centric workflow (everything under clients/{name}/)
• Automatic pathway detection (onboarding, module, skill, config, flex)
• New client name detection triggers onboarding
• Pre-flight input validation before skill execution
• Platform credential collection during onboarding
• Skill outputs saved as human-readable Markdown
• Client context loading (Claude sees existing data)
• One-liner install: curl ... | bash

v0.5.0 (January 26, 2026)

Production Foundation:

• Release policy (auto_deliver, auto_refine, human_review, blocked)
• Per-tenant budget tracking
• Skill templates with full contracts
• Execution modes (suggest/preview/execute)

v0.4.0 (January 25, 2026)

Research-Based Enhancements:

• SRAC evaluation framework
• TATS forecasting
• AI washing detection
• Multi-agent pipelines
• 70+ skills

---

License

Proprietary — MarketerHire Internal Use

---

Support

• Read this README
• Check CLAUDE.md for AI context
• Ask MH1 directly — it's designed to help