fix: revert --interactive to boolean True/False, update all surfaces

The display simplification changed --interactive to auto/on/off, but
the runtime consumers (selfy.py, media_loop.py, input_handlers.py)
all need a boolean — they gate input processing loops.

Decision: --interactive stays boolean (True/False) for the CLI and
runtime. The 'auto' concept (DM campaigns prompt, generative sims
don't) is handled internally by the simulation layer based on campaign
type, not exposed as a CLI flag value.

Changes:
- cli_parser.py: --interactive back to True/False (removed duplicate)
- cli_utils.py: validate as boolean
- cli.py: convert bool to on/off for sim display layer
- api.py: configure() docstring updated (on/off, not auto)
- display_simplification_plan.md: documented the decision
- test defaults updated

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: dennys246

docs/plans/display_simplification_plan.md

@@ -2,10 +2,12 @@
 
 **Goal:** Replace 4 overlapping output systems and 5+ display flags with two orthogonal flags that control what the user sees and how they interact. Default should be clean narrative with smart interactive behavior.
 
-**Status:** Planning  
+**Status:** Partially shipped (DisplayTier + sim_logger display system). `--interactive` reverted to boolean True/False — "auto" detection belongs in the simulation layer, not the CLI flag.  
 **Priority:** First post-publication UX improvement (v1.0.1)  
 **Estimated effort:** ~500 LOC across ~15 files
 
+> **Decision (2026-04-09):** `--interactive` stays as a boolean (True/False) controlling whether the runtime accepts user input. The "auto" concept (DM campaigns prompt, generative sims don't) is handled by the simulation layer internally based on campaign type — not exposed as a CLI flag value. The `InteractiveMode` enum in `sim_logger.py` still supports auto/on/off for the display system's internal use via `maxim.configure(interactive="on"/"off")`.
+
 ---
 
 ## The Problem Today

src/maxim/api.py

@@ -66,8 +66,8 @@ def configure(
         display: Output detail level.  ``"clean"`` (narrative only, default),
             ``"bio"`` (+ memory/learning annotations), ``"debug"``
             (+ full system traces).
-        interactive: Input mode.  ``"auto"`` (context-dependent, default),
-            ``"on"`` (always prompt), ``"off"`` (never prompt).
+        interactive: Input prompting.  ``"on"`` (prompt for user input),
+            ``"off"`` (headless — no prompts, use policy defaults).
 
     Example::
 

src/maxim/cli.py

@@ -710,9 +710,10 @@ def _force_exit_handler(signum, frame):
             if display_arg == "debug":
                 set_show_channels("all")
 
-            # Apply --interactive mode
-            interactive_arg = getattr(args, "interactive", "auto")
-            set_interactive_mode(interactive_arg)
+            # Apply --interactive mode to simulation display layer
+            # Runtime --interactive is boolean; sim display uses auto/on/off
+            _interactive_bool = bool(getattr(args, "interactive", True))
+            set_interactive_mode("on" if _interactive_bool else "off")
 
             # Apply --show channel filter (legacy, overrides if explicit)
             show_channels = getattr(args, "show_channels", None)
@@ -1307,7 +1308,7 @@ def get_internet_policy():
                 mode=mode,
                 audio=audio_enabled,
                 audio_len=float(getattr(args, "audio_len", 5.0) or 5.0),
-                interactive=getattr(args, "interactive", "auto") != "off",
+                interactive=bool(getattr(args, "interactive", True)),
             )
 
             if mode == "sleep":

src/maxim/cli_parser.py

@@ -25,10 +25,8 @@ def _build_parser() -> argparse.ArgumentParser:
     core.add_argument(
         "--interactive",
         type=str,
-        default="auto",
-        choices=["auto", "on", "off"],
-        help="Input mode: auto (DM campaigns prompt, generative sims don't, DEFAULT), "
-        "on (always prompt), off (never prompt — use policy defaults).",
+        default="true",
+        help="Enable interactive terminal input (True/False). Default: true.",
     )
     core.add_argument(
         "--verbosity",

src/maxim/cli_utils.py

@@ -32,10 +32,13 @@ def normalize_args(args: argparse.Namespace) -> None:
     else:
         raise SystemExit(f"Invalid --audio value: {args.audio!r} (expected True/False)")
 
-    # --interactive: "auto" (default), "on", or "off"
-    interactive_raw = str(getattr(args, "interactive", "auto")).strip().lower()
-    if interactive_raw not in ("auto", "on", "off"):
-        raise SystemExit(f"Invalid --interactive value: {args.interactive!r} (expected auto/on/off)")
+    interactive_raw = str(getattr(args, "interactive", "true")).strip().lower()
+    if interactive_raw in ("1", "true", "t", "yes", "y", "on"):
+        args.interactive = True
+    elif interactive_raw in ("0", "false", "f", "no", "n", "off"):
+        args.interactive = False
+    else:
+        raise SystemExit(f"Invalid --interactive value: {args.interactive!r} (expected True/False)")
 
     if str(getattr(args, "mode", "active")).strip().lower() == "sleep":
         args.audio = True
@@ -285,7 +288,7 @@ def reexec_with_mode(args: argparse.Namespace, *, mode: str) -> None:
         "--audio_len",
         str(float(getattr(args, "audio_len", 5.0) or 5.0)),
         "--interactive",
-        str(getattr(args, "interactive", "auto")),
+        "true" if bool(getattr(args, "interactive", True)) else "false",
     ]
     language_model = str(getattr(args, "language_model", "") or "").strip()
     if language_model:

tests/unit/test_cli_utils.py

@@ -38,7 +38,7 @@ class TestNormalizeArgs:
     def _make_args(self, **kwargs):
         defaults = {
             "audio": "true",
-            "interactive": "auto",
+            "interactive": "true",
             "mode": "active",
             "epochs": 0,
             "language_model": None,