Surface silent failures in installer + dashboard skill

install.sh — two silent-fail paths now fail loudly:

  1. The interactive "Experimental channel" radio-select did
     `exec bash <(curl -fsSL ...)`. If curl 404'd / 5xx'd / DNS hiccupped,
     `exec bash` ran an empty script and returned to the user's prompt
     with no output. Now we curl to a tmp file with `|| die` and check
     the file is non-empty before exec'ing.
  2. The git-clone fallback in setup-mcp (and the FORCE refresh path)
     swallowed errors. A failed re-clone left $REPO_DIR missing,
     downstream `uv pip install -e "$REPO_DIR/..."` would explode with
     an unrelated error. Now both fallback clones `|| die` with a clear
     "Could not clone branch '<X>' from <URL>" message.

databricks-aibi-dashboards SKILL.md — `--dataset-catalog` and
`--dataset-schema` are FLAG-ONLY (no JSON-body equivalent). The CLI
silently warns "unknown field" and proceeds, creating a broken
dashboard whose queries can't resolve catalog.schema.

Made the rule explicit in both the Quick Reference row and the Step 5
canonical comment block:

  --dataset-catalog / --dataset-schema : FLAG-ONLY (REQUIRED; CLI
      silently drops them if put in --json).
  parent_path : JSON-ONLY (no flag).
  display_name / warehouse_id / serialized_dashboard : either form.

Live-verified the silent-drop behavior on CLI v0.296.

Co-authored-by: Isaac

Author: Quentin Ambard

databricks-skills/databricks-aibi-dashboards/SKILL.md

@@ -17,7 +17,7 @@ A dashboard should be showing something relevant for a human, typically some KPI
 | List tables | `databricks experimental aitools tools query --warehouse WH "SHOW TABLES IN catalog.schema"` |
 | Get schema | `databricks experimental aitools tools discover-schema catalog.schema.table1 catalog.schema.table2` |
 | Test query | `databricks experimental aitools tools query --warehouse WH "SELECT..."` |
-| Create dashboard | `databricks lakeview create --display-name "X" --warehouse-id "WH" --dataset-catalog CATALOG --dataset-schema SCHEMA --serialized-dashboard "$(cat file.json)" --json '{"parent_path": "/Workspace/Users/<you>/path"}'` — `parent_path` is JSON-only (no flag); everything else stays as flags. Queries must use bare table names. |
+| Create dashboard | `databricks lakeview create --display-name "X" --warehouse-id "WH" --dataset-catalog CATALOG --dataset-schema SCHEMA --serialized-dashboard "$(cat file.json)" --json '{"parent_path": "/Workspace/Users/<you>/path"}'` — `--dataset-catalog` / `--dataset-schema` are **flag-only** (REQUIRED; CLI silently drops them if put in `--json`); `parent_path` is JSON-only (no flag). Queries must use bare table names. |
 | Update dashboard | `databricks lakeview update DASHBOARD_ID --serialized-dashboard "$(cat file.json)"` |
 | Publish | `databricks lakeview publish DASHBOARD_ID --warehouse-id WH` |
 | Delete | `databricks lakeview trash DASHBOARD_ID` |
@@ -120,11 +120,18 @@ After deploying, the same `lakeview` subcommands manage the dashboard's lifecycl
 
 ```bash
 # Deploy: creates the dashboard in the workspace and returns a dashboard ID.
-# Canonical form — everything is a flag EXCEPT parent_path (JSON-only, no flag;
-# without it the dashboard lands at /Users/<you>/<display-name>).
-# --dataset-catalog/--dataset-schema inject the catalog/schema into each saved
-# dataset; queries inside dashboard.json MUST use bare table names only (e.g.,
-# "FROM trips"), NOT "FROM schema.trips" or "FROM catalog.schema.trips".
+# Canonical form — MIX flags + --json. Each field has exactly ONE valid place:
+#   --dataset-catalog / --dataset-schema : FLAG-ONLY (REQUIRED — no JSON field).
+#       The CLI silently warns "unknown field" and drops them if put in --json,
+#       leaving every dataset query unable to resolve its catalog.schema.
+#   parent_path : JSON-ONLY (no flag). Without it, dashboard lands at
+#       /Users/<you>/<display-name>.
+#   display_name / warehouse_id / serialized_dashboard : either form works;
+#       prefer flags for readability.
+# Queries inside dashboard.json MUST use bare table names ("FROM trips", never
+# "FROM schema.trips" or "FROM catalog.schema.trips") — --dataset-catalog and
+# --dataset-schema only fill in missing parts, they do NOT rewrite hardcoded
+# prefixes.
 databricks lakeview create \
   --display-name "My Dashboard" \
   --warehouse-id "abc123def456" \

install.sh

@@ -1309,16 +1309,20 @@ check_deps() {
 setup_mcp() {
     step "Setting up MCP server"
 
-    # Clone or update repo
+    # Clone or update repo. Fallback clones must `|| die` — a silent failure
+    # leaves $REPO_DIR missing and downstream `uv pip install -e "$REPO_DIR/..."`
+    # explodes with an unrelated error.
     if [ -d "$REPO_DIR/.git" ]; then
         git -C "$REPO_DIR" fetch -q --depth 1 origin "$BRANCH" 2>/dev/null || true
         git -C "$REPO_DIR" reset --hard FETCH_HEAD 2>/dev/null || {
             rm -rf "$REPO_DIR"
-            git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR"
+            git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR" \
+                || die "Could not clone branch '$BRANCH' from $REPO_URL — check your network and that the branch exists."
         }
     else
         mkdir -p "$INSTALL_DIR"
-        git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR"
+        git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR" \
+            || die "Could not clone branch '$BRANCH' from $REPO_URL — check your network and that the branch exists."
     fi
     ok "Repository cloned ($BRANCH)"
 
@@ -2110,8 +2114,15 @@ prompt_channel() {
         [ "$INSTALL_SKILLS" = false ] && args="$args --mcp-only"
         [ "$BRANCH_EXPLICIT" = true ] && args="$args --branch $BRANCH"
 
-        # Download and execute the experimental installer
-        exec bash <(curl -fsSL "https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/experimental/install.sh") $args
+        # Download and execute the experimental installer.
+        # Download to a tmp file first so a curl failure (network/404/5xx) surfaces
+        # — `exec bash <(curl ...)` would silently exec an empty script otherwise.
+        local exp_installer
+        exp_installer=$(mktemp -t ai-dev-kit-exp-installer.XXXXXX) || die "Could not create temp file for experimental installer"
+        curl -fsSL "https://raw.githubusercontent.com/databricks-solutions/ai-dev-kit/experimental/install.sh" -o "$exp_installer" \
+            || die "Could not download experimental installer from GitHub. Check your network connection, then retry."
+        [ -s "$exp_installer" ] || die "Downloaded experimental installer is empty — GitHub may be having issues. Retry in a moment."
+        exec bash "$exp_installer" $args
     fi
 }
 
@@ -2305,14 +2316,16 @@ main() {
             git -C "$REPO_DIR" fetch -q --depth 1 origin "$BRANCH" 2>/dev/null || true
             git -C "$REPO_DIR" reset --hard FETCH_HEAD 2>/dev/null || {
                 rm -rf "$REPO_DIR"
-                git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR"
+                git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR" \
+                    || die "Could not clone branch '$BRANCH' from $REPO_URL — check your network and that the branch exists."
             }
             ok "Repository refreshed ($BRANCH)"
         fi
     else
         step "Downloading sources"
         mkdir -p "$INSTALL_DIR"
-        git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR"
+        git -c advice.detachedHead=false clone -q --depth 1 --branch "$BRANCH" "$REPO_URL" "$REPO_DIR" \
+            || die "Could not clone branch '$BRANCH' from $REPO_URL — check your network and that the branch exists."
         ok "Repository cloned ($BRANCH)"
     fi