Agent
Use this agent for any Marketing Mix Model (MMM) work — scoping and planning an MMM engagement, auditing and preparing data, designing model architecture and priors, building and fitting models, diagnosing convergence or attribution problems, interpreting contributions and ROAS, optimising budget, designing…
Agent
Specialist sub-agent for MMM diagnostics. Assesses convergence, fit, generalisation, baseline health, prior-to-posterior learning and attribution plausibility for a fitted run, and reports whether the results are safe to act on. Invoked by agent-mmm for review and debugging work.
Agent
Specialist sub-agent for the MMM iterative improvement loop. Runs the tournament of model variants, scores each run, and applies posterior-informed prior tightening to refine the best model. Invoked by agent-mmm when the task is improving or optimizing an existing MMM.
Agent
Specialist sub-agent for MMM model construction and fitting. Compiles a spec into a framework, generates priors, runs the prior-predictive → calibrate → fit → posterior-predictive pipeline, and saves run artefacts. Invoked by agent-mmm for build and fit work.
Agent
Specialist sub-agent for generating stakeholder-specific MMM reports. Produces CMO, CFO, Marketing Ops, and Data Science reports from fitted model results. Handles target-unit-aware framing (CPA vs ROAS, monetary vs acquisition). Invoked by agent-mmm when the task is report generation or results presentation.
Command
Run automated data quality audit on the dataset defined in spec.yaml. Produces audit.json and auditreport.md in ./mmm-workspace/audit/.
Command
Compile spec.yaml plus recommended priors into the target framework, validate the model structure, and report anything the framework cannot express.
Command
Run diagnostics on a completed MMM run — convergence, fit, generalisation, baseline health, prior-to-posterior learning, and attribution plausibility.
Command
Fit the MMM model — runs prior predictive check, MCMC sampling, and posterior predictive check. Attaches any lift-test constraints, then saves the model and metrics to ./mmm-workspace/runs/ /.
Command
Run the iterative improvement tournament — fits N model variants per round, scores them, and uses the winner's posterior to tighten priors for the next round. Persists leaderboard.
Command
Quick 5-question MMM project intake. Creates ./mmm-workspace/spec.yaml with sensible defaults. Run /mmm-intake for the full 25-question version.
Command
Full MMM project intake questionnaire (25 questions). Creates or updates ./mmm-workspace/spec.yaml. Covers company, target, channels, seasonality, controls, greenfield/brownfield, and multi-geo. Run /mmm-intake-quick for the 5-question fast version.
Command
Clean and shape the raw dataset for MMM — aggregate to the modelling period, make the calendar dense, impute missing values with an explicit rule per column, and add calendar features. Writes a prepared dataset plus a record of every change.
Command
Recommend external factor and control variable columns for your MMM based on industry, region, and detected data patterns.
Command
Generate Bayesian prior parameter recommendations for all channels in your MMM. Produces modelconfig.json and a prior audit report.
Command
Generate stakeholder reports for the best MMM run. Supports cmo, cfo, mops, ds, or all.
Command
Show the status of the current MMM workspace — what's been done, what's next.
Skill Claude CodeCodex
Exact pymc-marketing API reference for MMM code, verified against v1.1.0. Use when writing or reviewing pymc-marketing code, checking constructor signatures, method names, return types, import paths, or plotting/evaluation/optimisation calls. Also use when migrating 0.x code to 1.x, debugging ImportError or…
Skill Claude CodeCodex
Turning a fitted MMM into channel contributions, ROAS, CPA, marginal returns and response curves — correctly. Use when extracting contributions, computing return metrics with uncertainty, reading saturation curves, decomposing the target, comparing efficiency across channels, or diagnosing an implausible attribution…
Skill Claude CodeCodex
The baseline of an MMM — intercept, trend, seasonality and controls — and how it competes with media for variance. Use when the baseline is negative, implausibly large or small, when deciding between a fixed intercept, a linear trend, changepoints, knots or a time-varying intercept, when seasonality is absorbing…
Skill Claude CodeCodex
Budget allocation and scenario analysis from a fitted MMM. Use when optimising spend across channels, setting bounds and constraints, running sensitivity or what-if scenarios, comparing current versus optimal allocation, choosing between fixed-budget and target-return scenarios, or judging whether an optimiser's…
Skill Claude CodeCodex
Causal identification for MMM — what to control for, what never to control for, and why. Use when deciding which variables to include, when a channel's estimate looks structurally wrong, when handling mediators (search volume, site visits, brand metrics), colliders, funnel effects or reverse causality, when building a…
Skill Claude CodeCodex
How each type of marketing channel should be represented in an MMM — its role (paid media, reach-and-frequency, organic, non-media treatment, control), whether it is driven by spend or exposure, its carryover and saturation behaviour, and whether a ROAS is even defined for it. Use when deciding what a column means…
Skill Claude CodeCodex
Building the MMM dataset: sourcing, joining, aggregating, handling missing data, currency and inflation, fiscal calendars, taxonomy mapping, outliers and reproducible pipelines. Use when assembling data from platform exports and finance systems, deciding how to fill gaps, choosing daily vs weekly, aligning spend to…