A collection of interactive skills for AI coding assistants to guide users through building, visualizing, optimizing, and scenario planning with the Meridian Marketing Mix Modeling (MMM) framework.
These skills are designed for interactive AI coding environments, providing step-by-step guidance, smart defaults, and reproducible Python script generation across the entire MMM lifecycle.
Install Meridian skills directly into your workspace:
# Install all Meridian skills in your project
npx skills add google/meridian
# Or install a specific skill (e.g., model building)
npx skills add google/meridian --skill meridian-model-building
# Or install globally across all projects
npx skills add google/meridian --global --yes-
In-Repo Direct Discovery: When developing inside a clone of the
google/meridianrepository, Antigravity IDE automatically discovers and indexes all skills via.agents/skills.json. -
Plugin Installation:
agy plugin install https://github.com/google/meridian/tree/main/skills
Any AI assistant supporting workspace skill manifests or directory-based
instructions can load these skills by referencing the skills/ directory.
| Skill | Directory | Description |
|---|---|---|
meridian-doc-consultant |
meridian_doc_consultant |
Answers conceptual, parameter, and best practice questions about Meridian MMM using relevant documentation. |
meridian-model-building |
meridian_model_building |
Guides users through loading data, column mapping, ModelSpec configuration, EDA, model fitting, and saving. |
meridian-result-visualization |
meridian_result_visualization |
Loads fitted Meridian models and generates Model Results Summary and diagnostics visualizations. |
meridian-budget-optimization |
meridian_budget_optimization |
Loads fitted models to compute optimal budget allocation and target ROI spending scenarios. |
meridian-scenario-planner |
meridian_scenario_planner |
Generates Scenario Planner data, uploads to Google Sheets, and provides dashboard URLs. |
- Progressive Disclosure: The AI agent only indexes the skill's name and summary initially. Full instructions and templates are loaded on-demand when the skill is activated.
- Interactive Checkpoints: The agent pauses before critical decisions (column mapping, prior distributions, script execution) to present proposals and confirm user approval.
- Reproducible Python Scripts: Generated workflows produce clean,
standalone Python scripts runnable via
python3against your active Python environment.
- Bug Reports & Feedback: Please file issues or feature requests in the GitHub Issue Tracker.
- License: Apache 2.0 (see LICENSE for details).