MCNOS Skills

Open standards and deterministic tooling for governed AI marketing workflows.

What is MCNOS?

MCNOS is an AI marketing operations platform designed for international channels. It turns brand evidence into structured campaigns while keeping people in control of decisions that affect the brand or publish content.

The operating loop is:

Brand evidence -> Campaign plan -> Human approval -> Channel content
              -> Human approval -> Publish -> Measure -> Accepted learning

MCNOS separates portable skill instructions from execution authority. A skill can describe inputs, outputs, permissions, evidence, approval gates, and stop conditions. The hosted MCNOS runtime decides whether an allow-listed TypeScript handler may execute it. A skill package never gains publishing authority by itself.

What is in this repository?

This public repository contains the reusable skill layer:

• skills/mcnos-create-skill/ — the canonical generator for a governed MCNOS skill, usable as an Agent Skill in Codex.
• schemas/ — JSON Schema contracts for manifests, skill runs, approvals, evidence, channel content, campaign planning, production, publishing, learning, and site events.
• scripts/ — deterministic validation and reproducible packaging tools.
• test-vectors/ — a complete fictional en-US campaign contract flow.
• .github/workflows/validate.yml — CI validation, deterministic packages, SHA-256 checksums, and tagged artifact attestations.

The hosted SaaS runtime, official business skills, private channel policies, customer data, credentials, provider integrations, and deployment configuration are intentionally not included.

Skill structure

Every MCNOS skill follows the same portable layout:

skills/<mcnos-skill>/
├── SKILL.md
├── agents/
│   ├── openai.yaml
│   └── mcnos.yaml
├── references/
├── scripts/
└── assets/

SKILL.md contains the trigger, primary workflow, stop conditions, and direct links to supporting references. agents/mcnos.yaml is declarative metadata: it defines the skill version, handler name, schemas, permissions, approvals, cost class, and execution limits, but it is never executable authority.

Quick start

Requirements: Python 3.11 or newer.

git clone https://github.com/tengbot/mcnos-skills.git
cd mcnos-skills
python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install -r requirements-dev.txt

python3 scripts/validate_repo.py
python3 scripts/validate_contracts.py
python3 -m unittest discover -s tests -v

Create a new portable skill:

python3 skills/mcnos-create-skill/scripts/create_mcnos_skill.py \
  mcnos-example-skill \
  --path skills \
  --description "Create a governed example marketing artifact." \
  --handler campaign.example \
  --permission brand.read \
  --approval final_output

python3 skills/mcnos-create-skill/scripts/validate_mcnos_skill.py \
  skills/mcnos-example-skill

Build a reproducible package and checksum:

python3 scripts/package_skill.py skills/mcnos-example-skill

Governance model

• Source instructions and schemas are written in English; outputs are localized through explicit locale and market fields.
• Execution status and result quality are separate concepts.
• Factual claims carry evidence references.
• Publish-capable workflows require explicit approval IDs.
• Skill packages cannot contain credentials, customer data, browser profiles, or arbitrary SaaS executables.
• Production runtimes should consume an exact SemVer, checksum, and verified repository provenance.

Contributing and security

See CONTRIBUTING.md before proposing a schema or skill change. Please report security issues through the process in SECURITY.md, not a public issue.

License

Apache License 2.0. See LICENSE. MCNOS names and marks are not granted under the Apache license except as required to describe the origin of this work.