pymc-labsOfficial·13 Agent Skills Included

decision-lab

Parallel data science runs with convergence-checked, trustworthy reports

Runs data analyses multiple ways in parallel using different models and assumptions, then checks whether results converge before making recommendations. Packages prompts, domain skills, and locked Docker environments into reusable decision-packs for marketing mix modeling, forecasting, and Bayesian analysis. Eliminates confident-but-wrong reports by flagging disagreement, diagnosing failures, and suggesting experiments when data cannot support a conclusion.

178|13

↗

npx skills add pymc-labs/decision-lab --all -g -y

Available:

Tells the AI agent how the dlab CLI, decision-pack architecture, parallel subagents, and session monitoring work so it can run, scaffold, and debug analysis sessions correctly.

All Skills in This Repository (13)

Pure Emerald Level Indicators

📦 In Repo

pymc-labs

Create decision-pack Interactively

Guide users through interactive questions to create a dlab decision-pack configuration.

Design data science agent systems

Design data science agent systems with anti-fabrication protocols and retry mechanisms.

Create decision-pack Programmatically

Generate decision-packs for data science agents via Python code.

Analyze dlab session runs

Analyze dlab session logs and outputs for convergence, consistency, and errors.

TUI Design System

Standardize themes, layouts, typography, and color palettes for Textual TUI applications.

opencode

Generate code snippets and analyze code for issues using Python and AI models.

Informative Priors for MMM

Set informative priors for PyMC-Marketing MMM models using domain knowledge.

PyMC Samplers

Configure and troubleshoot PyMC sampling methods with NUTS, Metropolis, and Hamiltonian samplers.

PyMC Data Handling

PyMC Distributions

Manage, debug, and apply PyMC distributions across PyMC3 and latest versions.

PyMC-Marketing MMM

Fit Bayesian Generalized Additive Models for Marketing Mix Modeling with PyMC-Marketing.

event-forecasting

Generate probabilistic forecasts for future events using Bayesian models and survival analysis.

Official

Advanced

Frequently Asked Questions

FAQPage Schema

How to install decision-lab?▼

Run `npx skills add pymc-labs/decision-lab --all -g -y` in your terminal to install all skills in this suite globally.

What problem does decision-lab solve?▼

It stops agents from committing to the first analysis that runs by exploring multiple modeling approaches in parallel and checking whether they converge. If results conflict, it says so and recommends experiments instead of making unsupported recommendations.

What is a decision-pack?▼

A decision-pack is a directory bundling agent prompts, domain skills, tools, and a locked Docker environment for one analytical domain. You point it at your data with a single dlab command and get back reports, figures, and recommendations.

Can I monitor running analysis sessions?▼

Yes. Use `dlab connect` for a live terminal dashboard, `dlab timeline` for a Gantt chart with per-agent costs, or `dlab view` for a browser-based DAG visualization of the whole session.

Do I need Docker to use decision-lab?▼

Docker is the default for reproducible, locked environments, but you can pass --no-sandboxing to run locally and let the agent set up its own environment.