diff --git a/docs/PLAN.md b/docs/PLAN.md
index 5aaa55c..2ac84e0 100644
--- a/docs/PLAN.md
+++ b/docs/PLAN.md
@@ -386,14 +386,16 @@ program, click **Run** (or Ctrl+Return); each Run is one undo step.
 - **Deviations (documented)**: regexes use `QRegularExpression` (PCRE), not
   plan9 `regexp(7)`. Everyday patterns match identically; the one semantic
   divergence is leftmost-greedy vs sam's leftmost-longest (bites only on
-  overlapping alternation like `a|ab`), and `^`/`$` anchor buffer-wide not
-  per-line by default. katesam gains PCRE extras (`\d \w \b`, lookahead,
-  non-greedy, inline flags). Full user-facing treatment with verified examples
-  and workarounds in **`docs/SAM.md`**. Out of scope: multi-file menu
-  (`b B n D`), external file I/O (`e r w f`), `"re"` file-addressing, and sam's
-  own `u` (Kate's undo stack is the undo mechanism). The live GUI panel (tool
-  view + transaction apply) is not exercisable headless and is unverified on a
-  live display.
+  overlapping alternation like `a|ab`). `^`/`$` are per-line (compiled with
+  `MultilineOption`) and `.` does not cross newlines — both faithful to sam's
+  `regexp(7)` (BOL/EOL in `regexp.c`). PCRE also accepts extras sam lacks
+  (`\d \w \b`, lookahead, non-greedy, inline flags); the docs frame these as an
+  escape hatch, not a feature — prefer structural composition. Full user-facing
+  treatment with verified examples in **`docs/SAM.md`**. Out of scope: multi-file
+  menu (`b B n D`), external file I/O (`e r w f`), `"re"` file-addressing, and
+  sam's own `u` (Kate's undo stack is the undo mechanism). The live GUI panel
+  (tool view + transaction apply) is not exercisable headless and is unverified
+  on a live display.
 
 ## Build and test
 
diff --git a/docs/SAM.md b/docs/SAM.md
index 41d65f6..21269ee 100644
--- a/docs/SAM.md
+++ b/docs/SAM.md
@@ -15,10 +15,11 @@ Every example below is verified against the engine.
 - Everyday patterns behave **identically** to sam.
 - One real semantic difference: **PCRE is leftmost-greedy, sam is
   leftmost-longest.** It only bites on *overlapping alternation* (`a|ab`).
-- `^` and `$` anchor to the **whole buffer, not each line**, by default. This is
-  the most common surprise. Use a structural loop or the `(?m)` flag.
-- You *gain* PCRE features sam never had: `\d \w \s \b`, lookahead/lookbehind,
-  non-greedy `*?`, backreferences, inline flags `(?i)` `(?m)` `(?s)`.
+- `^` and `$` are **per-line** (match at every line boundary), the same as sam's
+  `regexp(7)`. `.` does not cross newlines, also the same as sam.
+- PCRE *also* offers things sam lacks (`\d \w \b`, lookahead, non-greedy, inline
+  flags). They work, but prefer sam-style structural composition (`x g v`) — see
+  "Extras" below.
 
 ---
 
@@ -39,46 +40,34 @@ silently produce a different edit in katesam. Non-overlapping alternation
 (`cat|dog`) and all non-alternation patterns are unaffected.
 
 Quantifier greediness is the same in both (`a.*c` on `axxcxxc` → whole string in
-both). katesam additionally offers non-greedy `.*?`:
-
-| program | input | result |
-|---|---|---|
-| `,s/a.*c/X/` | `axxcxxc` | `X` (greedy, same as sam) |
-| `,s/a.*?c/X/` | `axxcxxc` | `Xxxc` (non-greedy, PCRE bonus) |
+both). katesam additionally offers non-greedy `.*?` (a PCRE extra).
 
 ---
 
-## Anchors `^` / `$` are buffer-wide, not per-line
+## Anchors `^` / `$` are per-line — same as sam
 
-By default PCRE anchors `^`/`$` to the **start/end of the whole buffer**, so a
-global substitution touches only the first/last position — **not** every line:
+This is **not** a deviation. plan9 `regexp(7)` defines `^` as "the beginning of
+a line" and `$` as "the end of a line" (sam's `regexp.c`: `BOL` fires at offset
+0 or after a `\n`; `EOL` fires before a `\n`). katesam compiles every pattern
+with `QRegularExpression::MultilineOption`, so `^`/`$` match at every line
+boundary, exactly as sam does:
 
 | program | input | result |
 |---|---|---|
-| `,s/^/> /g` | `a⏎b⏎c` | `> a⏎b⏎c` (only line 1!) |
-| `,s/$/;/g` | `a⏎b⏎c` | `a⏎b⏎c;` (only the end!) |
+| `,s/^/> /g` | `a⏎b⏎c` | `> a⏎> b⏎> c` (every line) |
+| `,s/$/;/g` | `a⏎b⏎c` | `a;⏎b;⏎c;` (every line) |
 
-This differs from how people expect line-oriented edits to work. **Two correct
-ways** to act per line:
-
-**1. Structural (sam-idiomatic, preferred).** Loop over lines with `x`, then act
-on each line's dot. `.+` matches the non-newline run of each line:
+The sam-idiomatic structural form works too, and is preferable because it
+composes with guards and other loops:
 
 ```
-,x/.+/ i/> /        prefix every (non-empty) line with "> "   → > a⏎> b⏎> c
-,x/.+/ a/;/         append ";" to every line                  → a;⏎b;⏎c;
+,x/.+/ s/^/> /       prefix every (non-empty) line → > a⏎> b⏎> c
+,x/.+/ a/;/          append ";" to every line       → a;⏎b;⏎c;
 ```
 
-**2. PCRE multiline flag `(?m)`** (a bonus sam lacks). Makes `^`/`$` match at
-every line boundary:
-
-```
-,s/(?m)^/> /g       → > a⏎> b⏎> c
-,s/(?m)$/;/g        → a;⏎b;⏎c;
-```
-
-Prefer the structural form — it is the real sam way and composes with other
-loop/guard steps. Reach for `(?m)` for quick one-liners.
+`.` does **not** cross newlines (sam: a "character" is "any character but
+newline"), which is also PCRE's default. Line-structured descent like
+`,x/.*\n/ …` therefore behaves as expected.
 
 ---
 
@@ -95,36 +84,39 @@ loop/guard steps. Reach for `(?m)` for quick one-liners.
 
 > `.` not crossing newlines matches sam, so line-structured descent
 > (`,x/.*\n/ ...`) behaves as expected. Use `(?s)` if you *want* `.` to span
-> newlines (PCRE bonus).
+> newlines.
 
 ---
 
-## PCRE features sam never had (free upgrades)
+## Extras — available, but not the sam way
 
-All verified working in katesam:
+PCRE accepts syntax plan9 `regexp(7)` never had. It all works, but treat it as
+an **escape hatch, not a headline**: sam's power comes from *composing* simple
+patterns with `x`/`y`/`g`/`v`/`{}`, not from clever single regexes. Prefer
+structural composition; reach for these only when it is genuinely simpler.
 
-| feature | example | effect |
+| extra | example | note |
 |---|---|---|
-| shorthand classes | `,s/\d+/N/g` on `a12b345` | `aNbN` |
-| `\w+` in a loop | `,x/\w+/ c/W/` on `foo bar` | `W W` |
-| word boundary `\b` | `,s/\bfoo\b/X/g` on `foo foobar foo` | `X foobar X` |
-| backreference | `,s/(\w)\1/D/g` on `aabbc` | `DDc` |
-| lookahead | `,s/foo(?=bar)/X/g` on `foobar fooqux` | `Xbar fooqux` |
-| inline flags | `(?i)` ignore-case, `(?m)` multiline, `(?s)` dotall | per-pattern |
+| shorthand classes `\d \w \s` | `,s/\d+/N/g` | less typing than `[0-9]`; genuinely handy |
+| word boundary `\b` | `,s/\bfoo\b/X/g` | handy; sam would use `x/foo/` with context |
+| ignore-case `(?i)` | `,s/(?i)todo/DONE/g` | fills a real gap — sam has no case-insensitive match |
+| non-greedy `*?` | `,s/a.*?c/X/` | the structural `x/…/` loop is the sam alternative |
+| lookahead/behind | `,s/foo(?=bar)/X/` | un-sam; express context with `x`/`g`/`v` instead |
+| backreference in pattern | `,s/(\w)\1/D/g` | rarely the right tool |
 
-These make many edits *easier* than in real sam. The tradeoff is that a pattern
-written for katesam may not be portable back to sam.
+A pattern using these is **not portable back to sam**. If portability or
+staying in the structural idiom matters, avoid them.
 
 ---
 
 ## Practical guidance
 
 - **Order overlapping alternatives longest-first** (`\.tar\.gz|\.gz`, not the
-  reverse). The only silent divergence.
-- **For per-line edits, prefer `,x/.+/ …`** over `^`/`$`. If you use anchors
-  globally, remember they are buffer-wide, or add `(?m)`.
-- **Lean on PCRE extras** (`\b`, `\d`, lookahead) freely — they are reliable;
-  just know they are not sam-portable.
+  reverse). The only silent divergence from sam.
+- **`^`/`$` are per-line** — just like sam. For whole-line edits either anchor
+  directly (`,s/^/> /g`) or, more sam-idiomatically, loop (`,x/.+/ …`).
+- **Prefer structural composition** (`x g v {}`) over PCRE extras. The extras
+  work but pull you out of the sam idiom and are not sam-portable.
 - **Empty-match globals** (`s/x*/…/g`) advance one position per null match, as in
   sam — safe, but as always with `*`, double-check the result.
 
diff --git a/src/plugin/sampanel.cpp b/src/plugin/sampanel.cpp
index 354875b..3db0ff8 100644
--- a/src/plugin/sampanel.cpp
+++ b/src/plugin/sampanel.cpp
@@ -56,9 +56,8 @@ SamPanel::SamPanel(QWidget *parent)
     auto *hint = new QLabel(
         i18n("sam commands — e.g. <tt>,s/foo/bar/g</tt> or "
              "<tt>X/\\.cpp$/ ,s/old/new/g</tt> (project-wide). Ctrl+Return runs.<br/>"
-             "Regex is PCRE: <tt>^</tt>/<tt>$</tt> are buffer-wide (use "
-             "<tt>,x/.+/ …</tt> or <tt>(?m)</tt> per line); order overlapping "
-             "alternatives longest-first."),
+             "Regex is PCRE: <tt>^</tt>/<tt>$</tt> are per-line (like sam); order "
+             "overlapping alternatives longest-first. See docs/SAM.md."),
         this);
     hint->setWordWrap(true);
     hint->setTextFormat(Qt::RichText);
diff --git a/src/sam/samengine.cpp b/src/sam/samengine.cpp
index 468d3b9..ef25937 100644
--- a/src/sam/samengine.cpp
+++ b/src/sam/samengine.cpp
@@ -658,6 +658,13 @@ private:
 
     // Build a QRegularExpression from a sam pattern. The empty pattern reuses
     // the last compiled pattern (sam behaviour).
+    //
+    // Multiline is ON: plan9 regexp(7) defines ^ as "beginning of a line" and $
+    // as "end of a line" (sam regexp.c BOL = p==0 || prev=='\n'; EOL = next
+    // char is '\n'). That is exactly QRegularExpression::MultilineOption, so it
+    // is the faithful default, not an opt-in. '.' still does not cross newlines
+    // (sam: "the word character means any character but newline"), which is
+    // QRegularExpression's default (no DotMatchesEverythingOption).
     QRegularExpression compile(const QString &pat)
     {
         QString p = pat;
@@ -666,7 +673,7 @@ private:
         } else {
             m_lastRe = p;
         }
-        QRegularExpression re(p);
+        QRegularExpression re(p, QRegularExpression::MultilineOption);
         if (!re.isValid()) {
             fail(QStringLiteral("bad regexp: %1").arg(re.errorString()));
         }
diff --git a/src/sam/test_samengine.cpp b/src/sam/test_samengine.cpp
index 222edec..60b893f 100644
--- a/src/sam/test_samengine.cpp
+++ b/src/sam/test_samengine.cpp
@@ -53,6 +53,9 @@ private Q_SLOTS:
     void peelFileLoopY();
     void peelFileLoopNone();
     void errorOnNoMatch();
+    void caretIsPerLine();
+    void dollarIsPerLine();
+    void samAnchorIdiom();
 };
 
 void TestSamEngine::substituteFirst()
@@ -256,5 +259,28 @@ void TestSamEngine::errorOnNoMatch()
     QVERIFY(!r.error.isEmpty());
 }
 
+void TestSamEngine::caretIsPerLine()
+{
+    // plan9 regexp(7): ^ matches the beginning of EACH line (regexp.c BOL =
+    // p==0 || prev=='\n'). So a global ^ substitution prefixes every line.
+    QCOMPARE(runAll(QStringLiteral(",s/^/> /g"), QStringLiteral("a\nb\nc")),
+             QStringLiteral("> a\n> b\n> c"));
+}
+
+void TestSamEngine::dollarIsPerLine()
+{
+    // $ matches the end of EACH line (regexp.c EOL = next char is '\n').
+    QCOMPARE(runAll(QStringLiteral(",s/$/;/g"), QStringLiteral("a\nb\nc")),
+             QStringLiteral("a;\nb;\nc;"));
+}
+
+void TestSamEngine::samAnchorIdiom()
+{
+    // The canonical sam idiom: loop over lines and anchor within each line's
+    // dot. With per-line anchors this prefixes every line.
+    QCOMPARE(runAll(QStringLiteral(",x/.+/ s/^/> /"), QStringLiteral("a\nb\nc")),
+             QStringLiteral("> a\n> b\n> c"));
+}
+
 QTEST_MAIN(TestSamEngine)
 #include "test_samengine.moc"