diff --git a/CLAUDE.md b/CLAUDE.md
index caea5ac0..b66bdd36 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -25,9 +25,8 @@ npm run format:check # Check formatting without modifying
 ### Backend (FastAPI, `backend/`)
 
 ```bash
-uvicorn api.main:app --reload --port 6767  # API server on localhost:6767
-pip install -r requirements.txt            # Install dependencies
-streamlit run Home.py                      # Legacy Streamlit interface (not primary)
+uv sync                                            # Install dependencies
+uv run uvicorn api.main:app --reload --port 6767   # API server on localhost:6767
 uv run sqlfluff lint query/sql data_cleaning/sql   # Lint the Jinja SQL templates
 uv run sqlfluff fix query/sql data_cleaning/sql    # Auto-fix SQL lint issues
 ```
diff --git a/README.md b/README.md
index ddd6f188..ed84a72a 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,7 @@
 
 The **Vermont Data Collaborative** is an open-source web app that makes Vermont's public data easy to explore. It pulls together housing, demographic, zoning, flood hazard, soil suitability, parcel, and wastewater data from state and federal sources, and shows it through interactive maps, charts, and tables you can filter by town, county, or region. It's built for people who aren't data specialists, like Regional Planning Commissions, Select Boards, developers, reporters, and engaged residents. You can compare places, save charts to a working report and export it as a PDF, or download the raw data for your own analysis. The frontend runs on Next.js and the backend on FastAPI with DuckDB.
 
-New to the codebase? Read the [Codebase Primer](docs/Codebase_Primer.md); all other documentation is indexed in [docs/](docs/README.md).
+New to the codebase? Read the [Codebase Primer](docs/codebase_primer.md); all other documentation is indexed in [docs/](docs/README.md).
 
 ---
 
diff --git a/docs/Codebase_Primer.md b/docs/codebase_primer.md
similarity index 100%
rename from docs/Codebase_Primer.md
rename to docs/codebase_primer.md