diff --git a/AGENTS.md b/AGENTS.md
index 0eba5bc..9ea086b 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -88,8 +88,9 @@ Method 按以下三个小节展开，除非用户明确要求，不再改变这
    它不是部署时可获得的视觉特征，也不是性能上界。
 
 主结果表固定为 EasyCom 与 Ego4D 各四行：EgoSpeak A+V、Omni G、Omni G+A、
-Omni G+A+SVA，并保留十个 0.2--2.0 秒 horizon 与均值。尚未取得的 SVA/Ego4D
-分数用 `--` 占位，不插值、不把计划写成结果；pooled-V 与 TSVA 只进入视觉分析表。
+Omni G+A+SVA，并保留十个 0.2--2.0 秒 horizon 与均值。EasyCom SVA 曲线目前仅为
+用户提供、split/seed/checkpoint/评测协议待核对的探索值；Ego4D 未取得的分数仍用
+`--` 占位。不插值、不把计划写成结果；Pool 与 TSVA 只进入视觉分析表。
 
 ## 不可破坏的研究契约
 
diff --git a/manuscript/README.md b/manuscript/README.md
index 312a66e..1a28c5d 100644
--- a/manuscript/README.md
+++ b/manuscript/README.md
@@ -2,11 +2,14 @@
 
 ```text
 manuscript/
-  ICASSP2026_Paper_Templates/  # official template assets; keep unmodified
+  ICASSP2026_Paper_Templates/  # archived template assets
+  ICASSP2027_Paper_Templates/  # active template assets; keep unmodified
   main/                        # editable paper source and isolated build
+  targeted/                    # independent acceptance-target draft
 ```
 
-The active source is [`main/main.tex`](main/main.tex). Compile it with:
+Both manuscripts use the 2027 author kit through their `latexmkrc` files.
+Compile the main draft with:
 
 ```bash
 cd manuscript/main
@@ -14,6 +17,4 @@ make
 ```
 
 The generated PDF is `main/build/main.pdf`. Build artifacts are ignored by
-Git. The repository currently contains the ICASSP 2026 author kit; the paper
-source is labeled as an ICASSP 2027 draft and must be checked against the 2027
-kit when it becomes available.
+Git. The targeted draft builds the same way from [`targeted/`](targeted/).
diff --git a/manuscript/main/README.md b/manuscript/main/README.md
index 910ba80..8cb7c68 100644
--- a/manuscript/main/README.md
+++ b/manuscript/main/README.md
@@ -53,10 +53,10 @@ remain records of their dates. The new scoped decision is
 
 ## Build and historical scope
 
-This is the editable paper directory. It uses the unmodified official ICASSP
-2026 LaTeX assets stored in `../ICASSP2026_Paper_Templates/`. When the ICASSP
-2027 author kit is released, replace or add the official template and update
-`template_dir` in `latexmkrc` only after checking the new instructions.
+This is the editable paper directory. It uses the ICASSP 2027 LaTeX assets
+stored in `../ICASSP2027_Paper_Templates/`, selected by `latexmkrc`.
+The supplied 2027 `spconf.sty` and `IEEEbib.bst` match the 2026 copies;
+the 2027 `Template.tex` updates the conference guidance.
 
 Compile from this directory:
 
diff --git a/manuscript/main/latexmkrc b/manuscript/main/latexmkrc
index 860d59d..d14d1dc 100644
--- a/manuscript/main/latexmkrc
+++ b/manuscript/main/latexmkrc
@@ -1,5 +1,5 @@
 my $path_separator = $^O eq 'MSWin32' ? ';' : ':';
-my $template_dir = '../ICASSP2026_Paper_Templates';
+my $template_dir = '../ICASSP2027_Paper_Templates';
 
 $ENV{'TEXINPUTS'} =
     $template_dir . $path_separator . ($ENV{'TEXINPUTS'} // '');
diff --git a/manuscript/main/main.tex b/manuscript/main/main.tex
index 282cfb6..0051e6f 100644
--- a/manuscript/main/main.tex
+++ b/manuscript/main/main.tex
@@ -1,6 +1,6 @@
 % ICASSP 2027 working draft: results and mechanism claims are not yet frozen.
 % Four technical pages maximum; the optional fifth page contains references only.
-% Use the final author kit and resolve every TODO before submission.
+% Uses the repository's ICASSP 2027 author kit; resolve every TODO before submission.
 \documentclass{article}
 \usepackage{spconf}
 \usepackage{amsmath,amssymb,graphicx,hyperref,xcolor}
@@ -278,9 +278,8 @@ person boxes. Dashes indicate unmeasured values; locked rows await the final
 matched dense/boundary reruns.}
 \label{tab:main}
 \centering
-\scriptsize
-\setlength{\tabcolsep}{2.6pt}
-\resizebox{\textwidth}{!}{%
+\normalsize
+\setlength{\tabcolsep}{2pt}
 \begin{tabular}{@{}lccccccccccccc@{}}
 \hline
 & \multicolumn{10}{c}{Dense mAP at prediction horizon (s) $\uparrow$}
@@ -317,7 +316,7 @@ Locked $G+A+V$ (optional)
   & -- & -- & -- & -- & -- & -- & -- & -- & -- & --
   & -- & -- & --\\
 \hline
-\end{tabular}}
+\end{tabular}
 \end{table*}
 
 Alongside these models, a planned causal current-activity classifier repeats
@@ -356,7 +355,7 @@ a matched Qwen 1-Hz control; MiniCPM 5-Hz tail rows require GPU verification.
 
 \subsection{Streaming and Context Analysis}
 
-Table~\ref{tab:ablation} reserves a tail-refresh $\times$ local-audio
+Table~\ref{tab:ablation} reserves a tail-refresh/local-audio
 comparison. Completed-only states update at 2-s closures and are causally
 held between them; tail-aware states use the observed partial block at every
 cutoff. Enabled local audio remains 5~Hz. Common startup rows and valid pairs
@@ -369,7 +368,7 @@ AP (percent). Transfer reports $G+A-G$ deltas (points) after interface audits.
 Dashes are unmeasured, not zero.}
 \label{tab:ablation}
 \centering
-\scriptsize
+\normalsize
 \setlength{\tabcolsep}{3pt}
 \begin{tabular}{@{}lcccc@{}}
 \hline