[PATCH] llm: fix newline escape in the SKILL.md script skeleton

The skeleton's BEGIN and ERROR clauses use "\\n" where "\n" is meant.
In a D string literal that is an escaped backslash followed by 'n', so
printf() emits a stray backslash instead of a newline.

Agents copy the skeleton verbatim, and "\\n" is a valid literal, so the
'dtrace -e' check prescribed in the same file does not catch it.

Every other D snippet in the tree uses a single backslash.

Assisted-by: Kiro:claude-opus-4.8
Signed-off-by: Zhan Xusheng <[email protected]>
---
Not tested under DTrace; this is purely a string-literal escape.
Applies to devel (30c9c22).

One thing I left alone: the skeleton's ERROR clause prints probeprov/
probemod/probefunc/probename, but those describe the probe that is
currently firing, which inside an ERROR clause is dtrace:::ERROR itself
rather than the probe that faulted.
doc/userguide/reference/dtrace_providers_dtrace.md gives the culprit as
arg1 (its EPID) and the fault type as arg4, and
examples/language_features/001intro-ERROR-probe.d sidesteps this by
printing a static message.  Should the skeleton use arg1/arg4?

Also minor: the "Target one process" pattern needs -p PID or -c CMD,
which the suggested "sudo dtrace -s script.d" does not provide.

 llm/SKILL.md | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/llm/SKILL.md b/llm/SKILL.md
index e5e2f3e..a6ff502 100644
--- a/llm/SKILL.md
+++ b/llm/SKILL.md
@@ -66,7 +66,7 @@ probe-descriptions
 
 dtrace:::BEGIN
 {
-    printf("Tracing started...\\n");
+    printf("Tracing started...\n");
 }
 
 /* tracing clauses */
@@ -78,7 +78,7 @@ dtrace:::END
 
 dtrace:::ERROR
 {
-    printf("DTrace error at %s:%s:%s:%s\\n", probeprov, probemod, probefunc, probename);
+    printf("DTrace error at %s:%s:%s:%s\n", probeprov, probemod, probefunc, probename);
 }
 ```
 
-- 
2.43.0