feat(mac chat): f_θ default-ON in interactive chat + ADR 0015 update

- --chat now auto-enables --force-f-theta (full verifier/proposer/f_θ pipeline)
  unless --all-mlx-drafter (fast, f_θ-bypassed) or native baseline is chosen.
- mlx-kakeya-fused-chat-ftheta preset drops the explicit --force-f-theta to
  verify the DEFAULT-on behavior on the Mac.
- ADR 0015: new section 'Mac (MLX) interactive engine — full pipeline, f_θ
  default-ON' documenting the engine, the gemma-4 recall-irrelevant-but-runs
  caveat, the forensic timeline (f_θ S5-bypassed 2026-06-12 b3a04d0; proposer
  blocks=0 caught by 0a6fb19), and the measured result (f_theta_ran=TRUE 25
  sliding layers + proposer blocks>0).

Co-authored-by: FluffyAIcode <FluffyAIcode@users.noreply.github.com>

Author: cursoragent

docs/adr/0015-kakeya-attention-and-engine-substrate.md

@@ -93,6 +93,41 @@ full-attention fraction:
   *only* way to bound memory at full recall — and vLLM, having no restoration,
   **must keep full KV and cannot match it**. This is the engine's target regime.
 
+## Mac (MLX) interactive engine — full verifier/proposer/f_θ pipeline, f_θ default-ON
+
+The Apple-Silicon interactive CLI (`scripts/research/k3_integrated_niah_eval_mac.py
+--chat`) runs the **full Kakeya engine** — gemma-4 verifier (MLX) + **DFlash
+proposer** (fused spec-decode) + **f_θ K/V restoration** + **S5 bounded KV** — not
+verifier-only. It reuses the validated `fused_specdecode_generate_mlx_trim`
+per-turn sequence; the NIAH evidence loop is untouched.
+
+- **f_θ runs by default in chat.** `--force-f-theta` is auto-enabled in `--chat`
+  unless the fast all-MLX path (`--all-mlx-drafter`, f_θ bypassed) is explicitly
+  chosen. It bypasses the S5 native-prefill short-circuit so f_θ **executes** each
+  turn: it projects the proposer's hidden states → verifier K/V for the **25
+  sliding layers** and injects them.
+- **gemma-4 caveat (honest).** On gemma-4 those restored sliding-layer K/V are
+  **recall-irrelevant** — the 5 exact full-attention layers carry recall (the "S5
+  free lunch"), so f_θ's output is effectively *discarded by the recall path*. We
+  still run f_θ by default so the **full verifier/proposer/f_θ pipeline is
+  exercised end-to-end**; on **full-attention models** the same f_θ path is
+  load-bearing (it is the only way to bound memory at full recall).
+- **Forensic — when f_θ stopped running.** f_θ was silently bypassed under
+  `--s5-exact-full-attn` on **2026-06-12** by the *"Optimize MLX adaptive S5
+  native smoke path"* commits (`b3a04d0` / `1f6e58c`), which made
+  `build_restoration` short-circuit to `{}` under S5; the same *"adaptive S5
+  native"* path also let the proposer go to `blocks=0` while keeping the fused
+  label — caught by the evidence gate (`0a6fb19`, *"enforce PR #109 review
+  constraints"*) which added `--force-fused-specdecode`. Both squashed into main
+  via #117. f_θ remained S5-bypassed until `--force-f-theta` (this ADR's change)
+  made it default-on in the interactive chat.
+- **Measured (Mac M4, via the git-bus bridge).** Both chat turns:
+  **`f_theta_ran=TRUE`** restoring the **25 sliding layers** + **proposer
+  `blocks=2/4`, `mean_accept_len=4.0/3.5`** + correct answers ("Paris"; "red,
+  yellow, and blue") + natural `<end_of_turn>` stop + bounded resident KV
+  (12–18 MB). Torch-bridge path is slow (~0.5–6 tok/s); the all-MLX path
+  (proposer-only) is the fast option.
+
 ## Feasibility probes so far (informed the design — NOT the product)
 
 These ran on the eager-transformers research bench; they validate correctness and

inference_engine/bridge/manifest.py

@@ -729,7 +729,7 @@ def _harness_preset(
                     "--verifier-path", "${ENV:KAKEYA_MAC_VERIFIER_PATH}",
                     "--drafter-id", "${ENV:KAKEYA_MAC_DRAFTER_ID}",
                     "--f-theta-dir", "${ENV:KAKEYA_MAC_FTHETA_DIR}",
-                    "--s5-exact-full-attn", "--fused-specdecode", "--force-f-theta",
+                    "--s5-exact-full-attn", "--fused-specdecode",
                     "--sink-size", "4", "--window-size", "64",
                     "--block-size", "{block_size}",
                     "--max-new-tokens", "{max_new_tokens}",

scripts/research/k3_integrated_niah_eval_mac.py

@@ -262,6 +262,15 @@ def main() -> int:
         args.fused_specdecode = True
         args.force_fused_specdecode = True
     adaptive_s5_native = args.native_baseline_bypass
+    # Interactive chat runs the FULL verifier/proposer/f_θ pipeline by DEFAULT:
+    # f_θ executes each turn (torch drafter + f_θ) unless the fast all-MLX path
+    # (--all-mlx-drafter, f_θ bypassed) or the native baseline is explicitly chosen.
+    if args.chat and not args.all_mlx_drafter and not args.native_baseline_bypass:
+        if not args.force_f_theta:
+            print("[chat] f_θ default-ON for interactive chat (torch drafter + f_θ); "
+                  "pass --all-mlx-drafter for the fast f_θ-bypassed path.",
+                  file=sys.stderr, flush=True)
+        args.force_f_theta = True
     if args.all_mlx_drafter and not args.s5_exact_full_attn:
         raise SystemExit(
             "--all-mlx-drafter requires --s5-exact-full-attn: the all-MLX "

tests/inference_engine/bridge/test_manifest.py

@@ -156,8 +156,9 @@ def test_mlx_kakeya_fused_chat_ftheta_preset_runs_f_theta_path():
         params={"max_new_tokens": "32", "block_size": "4"}))
     (argv,) = build_commands(request, HARNESS_ENV)
     assert argv[1].endswith("k3_integrated_niah_eval_mac.py")
-    # torch drafter + f_θ path: --force-f-theta, and NOT --all-mlx-drafter
-    assert "--force-f-theta" in argv
+    # torch drafter + f_θ path: --chat WITHOUT --all-mlx-drafter → f_θ default-ON
+    # (no explicit --force-f-theta needed; the harness enables it for chat).
+    assert "--chat" in argv
     assert "--fused-specdecode" in argv
     assert "--all-mlx-drafter" not in argv
     assert HARNESS_ENV["KAKEYA_MAC_FTHETA_DIR"] in argv