HOW-TO-USE.md

@@ -484,7 +484,8 @@ the way they did: one that had Full Permissions on becomes **Bypass**, one that
 **Auto** sits between Accept Edits and Bypass: Claude Code's own classifier reviews each action,
 lets routine work through and blocks things that look risky (such as destructive or
 exfiltrating commands) — no prompts either way. Whether it is available depends on your Claude
-Code account and model; see Claude Code's documentation on permission modes.
+Code account, model and backend (local and OpenAI-compatible backends won't qualify). When it
+isn't available, Claude Code quietly starts in its normal prompting mode instead.
 
 ### When a change takes effect
 
@@ -502,7 +503,8 @@ Code account and model; see Claude Code's documentation on permission modes.
 > Scheduled tasks run headless (`claude -p`) and cannot answer a permission prompt. In any mode
 > other than **Auto** or **Bypass**, a task may simply stop early when Claude Code asks for
 > approval. In **Auto**, actions the classifier blocks are denied and the run carries on without
-> them. Its run log records which mode it used.
+> them — but if Auto isn't available for the project's model or backend, Claude Code falls back
+> to prompting and the task can stall the same way. Its run log records which mode it used.
 
 ---
 
@@ -1357,7 +1359,7 @@ with `--dangerously-skip-permissions`. Because the mode travels into the contain
 environment variable, **stop and start the project** after changing it for the scheduler to see the
 change. Remember that a headless run cannot answer a permission prompt, so in any mode other than
 **Auto** or **Bypass** a task may stop early when Claude Code asks for approval (in Auto, blocked
-actions are denied instead); the run log records the mode
+actions are denied instead, unless Auto is unavailable and Claude Code falls back to prompting); the run log records the mode
 that was used.
 
 ### Creating Tasks

app/src/components/projects/home/TaskEditorModal.test.tsx

@@ -165,10 +165,11 @@ describe("TaskEditorModal", () => {
     expect(screen.queryByText(/cannot answer a permission prompt/i)).toBeNull();
   });
 
-  it("tells Auto mode that blocked actions are denied, not prompted", async () => {
+  it("tells Auto mode that blocked actions are denied, and warns of the fallback", async () => {
     await renderEditor(null, { ...baseProject, permission_mode: "auto" });
     expect(screen.queryByText(/cannot answer a permission prompt/i)).toBeNull();
     expect(screen.getByText(/blocks are denied/i)).toBeInTheDocument();
+    expect(screen.getByText(/falls back to prompting/i)).toBeInTheDocument();
   });
 
   it("spells out the stall risk in any non-Bypass mode", async () => {

app/src/components/projects/home/TaskEditorModal.tsx

@@ -302,9 +302,10 @@ export default function TaskEditorModal({ project, task, onClose, onSaved }: Pro
             <strong className="text-[var(--text-primary)]">{modeLabel}</strong>).
           </p>
           {mode === "auto" && (
-            <p className="text-xs text-[var(--text-secondary)]">
+            <p className="text-xs text-[var(--warning)]">
-              In Auto mode nothing prompts: actions the safety classifier blocks are denied and the
+              In Auto mode, actions the safety classifier blocks are denied and the run carries on
-              run carries on without them, so check the log if a task seems to have skipped a step.
+              without them. If Auto isn&rsquo;t available for this project&rsquo;s model or backend,
+              Claude Code falls back to prompting and the task may stall.
             </p>
           )}
           {mode !== "bypass" && mode !== "auto" && (

container/triple-c-task-runner

@@ -61,9 +61,10 @@ TASK_TYPE=$(jq -r '.type' "$TASK_FILE")
 # project's permission setting. Keep this mapping in sync with
 # PermissionMode::cli_args() in app/src-tauri/src/models/project.rs.
 # NOTE: headless `claude -p` runs cannot answer a permission prompt, so any
-# mode other than "bypass" or "auto" means the task may stop early when Claude
+# mode other than "bypass" means the task may stop early when Claude Code asks
-# Code asks for permission. In "auto", actions the classifier blocks are
+# for permission. In "auto", classifier-blocked actions are denied instead of
-# denied rather than prompted, so the run continues without them. Unset or unrecognized values pass no flag (Claude's default).
+# prompted, but if auto mode is unavailable for the session's model/backend,
+# Claude Code falls back to prompting and the same stall applies. Unset or unrecognized values pass no flag (Claude's default).
 PERMISSION_ARGS=()
 case "${TRIPLE_C_PERMISSION_MODE:-}" in
     plan)        PERMISSION_ARGS=(--permission-mode plan) ;;