pymc-labs · drbenvincent · May 25, 2026 · May 25, 2026 · May 25, 2026 · May 25, 2026
diff --git a/.cursor/commands/causalpy_demos.md b/.cursor/commands/causalpy_demos.md
@@ -1,4 +1,3 @@
 # Causal Demos
 
-This command points to the `loading-datasets` Skill in `.github/skills/loading-datasets/`.
-Use that Skill for dataset discovery and example usage.
+Use `causalpy.load_data(name)` to load example datasets. See the `load_data()` docstring for available dataset names.
diff --git a/.cursor/commands/causalpy_estimators.md b/.cursor/commands/causalpy_estimators.md
@@ -1,4 +1,4 @@
 # Causal Estimators
 
 This command points to the `performing-causal-analysis` Skill in
-`.github/skills/performing-causal-analysis/` for estimation, summaries, and plots.
+`causalpy/skills/performing-causal-analysis/` for estimation, summaries, and plots.
diff --git a/.cursor/commands/causalpy_extras.md b/.cursor/commands/causalpy_extras.md
@@ -1,4 +1,4 @@
 # Causal Extras
 
 This command points to the `running-placebo-analysis` Skill in
-`.github/skills/running-placebo-analysis/` for the placebo-in-time workflow.
+`causalpy/skills/running-placebo-analysis/` for the placebo-in-time workflow.
diff --git a/.cursor/commands/causalpy_methods.md b/.cursor/commands/causalpy_methods.md
@@ -1,6 +1,6 @@
 # Causal Methods
 
-This command points to Skills in `.github/skills/`:
+This command points to Skills in `causalpy/skills/`:
 
 - `designing-experiments` for method selection
 - `performing-causal-analysis` for method usage and references
diff --git a/.cursor/commands/causalpy_research.md b/.cursor/commands/causalpy_research.md
@@ -1,4 +1,4 @@
 # Causal Research
 
 This command points to the `designing-experiments` Skill in
-`.github/skills/designing-experiments/` for method selection guidance.
+`causalpy/skills/designing-experiments/` for method selection guidance.
diff --git a/.github/skills/loading-datasets/SKILL.md b/.github/skills/loading-datasets/SKILL.md
diff --git a/.gitignore b/.gitignore
@@ -14,7 +14,6 @@ dist/
 docs/build/
 docs/jupyter_execute/
 docs/source/api/generated/
-
 .cursor/plans
 .marimo/
 

diff --git a/AGENTS.md b/AGENTS.md
@@ -100,7 +100,9 @@ creation, bug reports, and issue evaluation workflows.
 
 ## Skills Location
 
-Canonical skills live in `.github/skills/`. The `.claude/skills` and
-`.cursor/skills` paths are symlinks to that directory. On Windows, symlink
-support may require Developer Mode or elevated permissions; if symlinks are not
-available, mirror `.github/skills/` into those locations and keep them in sync.
+Skills are split into two categories with separate homes:
+
+- **Developer skills** live in `.github/skills/` and are auto-discovered by agents working on the repo via the `.cursor/skills`, `.claude/skills`, and `.agents/skills` symlinks. These cover environment setup, PR workflows, issue triage, and other maintainer tasks.
+- **User skills** live in `causalpy/skills/` inside the source tree. They teach AI agents how to use CausalPy for causal inference tasks. They are **not** symlinked into the auto-discovery paths — a developer agent should not see experiment-design skills mixed in with PR review skills. User skills are distributed via [Decision AI Hub](https://hub.decision.ai).
+
+On Windows, symlink support for developer skills may require Developer Mode or elevated permissions; if symlinks are not available, mirror `.github/skills/` into `.cursor/skills/`, `.claude/skills/`, and `.agents/skills/` and keep them in sync.
diff --git a/README.md b/README.md
@@ -61,6 +61,10 @@ Alternatively, if you want the very latest version of the package you can instal
 pip install git+https://github.com/pymc-labs/CausalPy.git
 ```
 
+## AI Agent Skills
+
+CausalPy includes agent skills that teach AI coding assistants how to use the library for causal inference. Skills are available via [Decision AI Hub](https://hub.decision.ai) and live in `causalpy/skills/` in the source tree.
+
 ## Quickstart
 
 ```python

diff --git a/causalpy/skills/README.md b/causalpy/skills/README.md
@@ -0,0 +1,13 @@
+# User-Facing Agent Skills
+
+Markdown skills in this directory teach AI coding agents how to use CausalPy for causal inference tasks. They are distributed to end users via [Decision AI Hub](https://hub.decision.ai).
+
+Developer-focused skills (environment setup, PR workflows, testing conventions, etc.) live in `.github/skills/` and are auto-discovered in-repo via platform symlinks. They are **not** included here.
+
+## Layout
+
+| Path | Purpose |
+|------|---------|
+| `designing-experiments/` | Choosing the right quasi-experimental method |
+| `performing-causal-analysis/` | Fitting models, estimating impacts, plotting results |
+| `running-placebo-analysis/` | Placebo-in-time sensitivity checks |
diff --git a/causalpy/skills/__init__.py b/causalpy/skills/__init__.py
@@ -0,0 +1,21 @@
+#   Copyright 2026 - 2026 The PyMC Labs Developers
+#
+#   Licensed under the Apache License, Version 2.0 (the "License");
+#   you may not use this file except in compliance with the License.
+#   You may obtain a copy of the License at
+#
+#       http://www.apache.org/licenses/LICENSE-2.0
+#
+#   Unless required by applicable law or agreed to in writing, software
+#   distributed under the License is distributed on an "AS IS" BASIS,
+#   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#   See the License for the specific language governing permissions and
+#   limitations under the License.
+"""CausalPy agent skills for AI-assisted causal inference.
+
+User-facing skills that teach AI agents how to use CausalPy.
+Distributed via Decision AI Hub (https://hub.decision.ai).
+
+Developer skills live in ``.github/skills/`` and are auto-discovered
+in-repo via platform symlinks; they are **not** included here.
+"""
diff --git a/...hub/skills/designing-experiments/SKILL.md → ...lpy/skills/designing-experiments/SKILL.md b/...hub/skills/designing-experiments/SKILL.md → ...lpy/skills/designing-experiments/SKILL.md
diff --git a/...kills/performing-causal-analysis/SKILL.md → ...kills/performing-causal-analysis/SKILL.md b/...kills/performing-causal-analysis/SKILL.md → ...kills/performing-causal-analysis/SKILL.md
@@ -26,3 +26,4 @@ Detailed usage for specific methods:
 *   [Difference-in-Differences](reference/diff_in_diff.md)
 *   [Interrupted Time Series](reference/interrupted_time_series.md)
 *   [Synthetic Control](reference/synthetic_control.md)
+*   [Regression Discontinuity](reference/regression_discontinuity.md)
diff --git a/...causal-analysis/reference/diff_in_diff.md → ...causal-analysis/reference/diff_in_diff.md b/...causal-analysis/reference/diff_in_diff.md → ...causal-analysis/reference/diff_in_diff.md
diff --git a/...ysis/reference/interrupted_time_series.md → ...ysis/reference/interrupted_time_series.md b/...ysis/reference/interrupted_time_series.md → ...ysis/reference/interrupted_time_series.md
diff --git a/causalpy/skills/performing-causal-analysis/reference/regression_discontinuity.md b/causalpy/skills/performing-causal-analysis/reference/regression_discontinuity.md
@@ -0,0 +1,57 @@
+# Causal Regression Discontinuity (RD)
+
+Regression Discontinuity exploits a cutoff or threshold in an assignment variable to identify causal effects. Units just above and below the threshold are compared to estimate the treatment effect at the discontinuity.
+
+## Class: `RegressionDiscontinuity`
+
+```python
+causalpy.experiments.RegressionDiscontinuity(
+    data,
+    formula,
+    treatment_threshold,
+    model=None,
+    running_variable_name="x",
+    epsilon=0.001,
+    bandwidth=np.inf,
+    donut_hole=0.0,
+    **kwargs
+)
+```
+
+### Parameters
+
+*   **`data`** (`pd.DataFrame`): Input dataframe.
+*   **`formula`** (`str`): Statistical formula (e.g., `"y ~ 1 + x + treated + x:treated"`).
+*   **`treatment_threshold`** (`float`): The cutoff value of the running variable where treatment is assigned.
+*   **`model`**: A PyMC model (e.g., `cp.pymc_models.LinearRegression`) or a Scikit-Learn Regressor.
+*   **`running_variable_name`** (`str`): Column name of the running variable. Default is `"x"`.
+*   **`epsilon`** (`float`): Small offset above/below the threshold for evaluating the causal impact. Default is `0.001`.
+*   **`bandwidth`** (`float`): Data outside this distance from the threshold is excluded from fitting. Default is `np.inf` (use all data).
+*   **`donut_hole`** (`float`): Observations within this distance from the threshold are excluded from fitting (robustness check). Default is `0.0`.
+
+### How it Works
+
+1.  **Fit**: Model is trained on data within the bandwidth, optionally excluding the donut hole.
+2.  **Predict**: Counterfactual predicted at the threshold by evaluating the model just above and just below.
+3.  **Impact**: The causal effect is the discontinuous jump in the outcome at the threshold.
+
+### Example
+
+```python
+import causalpy as cp
+import causalpy.pymc_models as cp_pymc
+
+df = cp.load_data("drinking")
+df = df.rename(columns={"agecell": "age"}).assign(treated=lambda d: d.age > 21)
+
+result = cp.RegressionDiscontinuity(
+    df,
+    formula="all ~ 1 + age + treated + age:treated",
+    running_variable_name="age",
+    model=cp_pymc.LinearRegression(),
+    treatment_threshold=21,
+)
+
+result.summary()
+result.plot()
+```
diff --git a/...l-analysis/reference/synthetic_control.md → ...l-analysis/reference/synthetic_control.md b/...l-analysis/reference/synthetic_control.md → ...l-analysis/reference/synthetic_control.md
diff --git a/.../skills/running-placebo-analysis/SKILL.md → .../skills/running-placebo-analysis/SKILL.md b/.../skills/running-placebo-analysis/SKILL.md → .../skills/running-placebo-analysis/SKILL.md
diff --git a/...ebo-analysis/reference/placebo_in_time.md → ...ebo-analysis/reference/placebo_in_time.md b/...ebo-analysis/reference/placebo_in_time.md → ...ebo-analysis/reference/placebo_in_time.md