Regex Explainer

/^\[(\d{2}:\d{2})\]\s(ERROR|WARN)\s(\w{2,5}):/

^ — Matches the start of the text — nothing may come before this point.
\[ — \[ matches [ itself — the backslash only escapes it.
(\d{2}:\d{2}) — Capturing group 1 — what it matches is kept and can be referred to later.
  \d{2} — Repeats the item before it exactly 2 times.
    \d — \d matches one digit, 0 to 9.
  : — Matches : — exactly that one character, code point 58.
  \d{2} — Repeats the item before it exactly 2 times.
    \d — \d matches one digit, 0 to 9.
\] — \] matches ] itself — the backslash only escapes it.
\s — \s matches one whitespace character — space, tab or line break.
(ERROR|WARN) — Capturing group 2 — what it matches is kept and can be referred to later.
  ERROR|WARN — Tries 2 alternatives in order and takes the first one that matches.
    ERROR — Alternative 1: the parts listed under it
      ERROR — Matches the text ERROR exactly, in that order — 5 characters.
    WARN — Alternative 2: the parts listed under it
      WARN — Matches the text WARN exactly, in that order — 4 characters.
\s — \s matches one whitespace character — space, tab or line break.
(\w{2,5}) — Capturing group 3 — what it matches is kept and can be referred to later.
  \w{2,5} — Repeats the item before it between 2 and 5 times.
    \w — \w matches one word character — letter, digit or underscore.
: — Matches : — exactly that one character, code point 58.

Three numbered captures, and the indentation is the pattern's own nesting: the two \d{2} lines sit inside group 1 because that is where those characters are. Groups are counted the way the JavaScript engine counts them.

/^(?=.{3,})(?<name>\p{L}+)-(?<id>\d+)$/u

^ — Matches the start of the text — nothing may come before this point.
(?=.{3,}) — Looks ahead: what follows must match this, and none of it is consumed.
  .{3,} — Repeats the item before it at least 3 times, with no upper limit.
    . — Matches any single character except a line break.
(?<name>\p{L}+) — Named capturing group name, which is also group 1 — what it matches is kept under that name.
  \p{L}+ — Repeats the item before it one or more times.
    \p{L} — \p{L} matches one character that has the Unicode property L.
- — Matches - — exactly that one character, code point 45.
(?<id>\d+) — Named capturing group id, which is also group 2 — what it matches is kept under that name.
  \d+ — Repeats the item before it one or more times.
    \d — \d matches one digit, 0 to 9.
$ — Matches the end of the text — nothing may come after this point.

The lookahead consumes nothing: it only asserts that three characters are available before the named part starts. Drop the u flag and the reading changes — \p{L} stops being a Unicode property, and a note says so rather than letting it quietly match four literal characters.

/(?>a)b/

Position 1: '(?>' is an atomic group, which exists in PCRE, Perl and .NET but not in JavaScript — JavaScript refuses it; use '(?:' and accept that this group can backtrack.
1 | (?>a)b
  | ^

EDGE CASE. An atomic group is ordinary syntax in PCRE, Perl and .NET, and JavaScript has none — so the answer is a refusal naming the dialect and the offset, not a guess at what you meant. The caret frame points at the character the reader stopped on.