diff --git a/.claude/CLAUDE.md b/.claude/CLAUDE.md
index 75f12b7..7601575 100644
--- a/.claude/CLAUDE.md
+++ b/.claude/CLAUDE.md
@@ -74,22 +74,28 @@ DUCKDB_PATH=local.duckdb SERVING_DUCKDB_PATH=analytics.duckdb \
 
 ```
 
-## Production queries
+## Production operations
 
-Use `scripts/prod_query.py` to query the production DuckDB over SSH. **Always prefer this over raw SSH commands** — it handles escaping, enforces read-only, and blocks mutation keywords.
+Use `scripts/prod.py` for all prod server operations over SSH. **Always prefer this over raw SSH commands** — it handles escaping, timeouts, and streaming.
 
 ```bash
 # Query analytics.duckdb (serving tables — default)
-uv run python scripts/prod_query.py "SELECT COUNT(*) FROM serving.location_profiles"
+uv run python scripts/prod.py query "SELECT COUNT(*) FROM serving.location_profiles"
 
 # Query lakehouse.duckdb (foundation/staging tables)
-uv run python scripts/prod_query.py --db lakehouse "SELECT * FROM foundation.dim_countries LIMIT 5"
+uv run python scripts/prod.py query --db lakehouse "SELECT * FROM foundation.dim_countries LIMIT 5"
 
 # JSON output
-uv run python scripts/prod_query.py --json "SELECT COUNT(*) FROM serving.location_profiles"
+uv run python scripts/prod.py query --json "SELECT COUNT(*) FROM serving.location_profiles"
 
-# Limit rows (default 500)
-uv run python scripts/prod_query.py --max-rows 1000 "SELECT ..."
+# Pipeline operations
+uv run python scripts/prod.py sqlmesh-plan --dry-run   # preview only
+uv run python scripts/prod.py sqlmesh-plan              # plan + auto-apply
+uv run python scripts/prod.py export                    # export serving tables
+uv run python scripts/prod.py status                    # supervisor status
+uv run python scripts/prod.py logs                      # last 100 log lines
+uv run python scripts/prod.py logs -f                   # follow logs
+uv run python scripts/prod.py deploy                    # blue/green deploy
 ```
 
 ## Architecture documentation