MCP server for SIMBA — connect AI assistants (Claude, Cursor) to Bayesian Marketing Mix Models. Upload data, build models, optimize budgets, and run scenarios through natural language.
This repository also configures its own agents. See what simba-mcp tells them →
Skill Claude CodeCodex
Build a Simba MMM from uploaded or pipeline data. Use for data preparation, model creation and fit recovery; use results guidance for existing-model questions.
Skill Claude CodeCodex
Set up Simba budget optimisation and inspect saved optimiser or scenario runs. Use for channel bounds, laydown, margins and decision-versus-comparison interpretation.
Skill Claude CodeCodex
Construct Simba media or control prior overrides. Use before changing carryover, saturation anchors, effect-at-average coordinates or explicit control transforms.
Skill Claude CodeCodex
Build and inspect Simba VAR models and linked long-run MMM results. Use for long-term effects, VAR fit recovery and long-run rollup interpretation.
Skill Claude CodeCodex
Read saved Simba results, ROI and diagnostics.
Skill Claude CodeCodex
Author, edit and review Simba Studies, recipes, policies and runs. Use for immutable publication, launch recovery, diagnostics and human handoff.
Skill Claude CodeCodex
Inspect campaign facts, mapping, incremental evidence, conditional budgets and experiment screening.
Skill Claude CodeCodex
Report actual KPI and spend from an existing stored dataset, preserving declared roles and source identity.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: