KM n-grams
辛気臭い	853
辛気くさい	629
心気臭い	6
心気くさい	2
辛気	1966
心気

The IPADIC-based parsing in the Google n-grams is a bit wonky in places. The KM ones, albeit on a smaller corpus, are more in the correct proportions.

@@ -24,2 +23,0 @@
-<xref type="see" seq="1720140">辛気臭い</xref>
-<xref type="see" seq="1720140">辛気臭い・1</xref>

daijs has no entry at all for 心気

daijs:
辛気臭い, entry, kanji for is: 辛気（臭い） (i.e くさい kanji optional).

daijs has an entry with two senses for 辛気 with two senses, and 辛気臭い is given as an extra headword within that.  Both, as noted earlier, are marked as dialects (we have marked 辛気臭い as kansai-ben).  Both senses appear to me to be variants of [2]:
[1] （もと関西方言）気がくさくさして、めいるようすだ。暗い。「ー話」
[2] （関西・中部・中国方言) いらだたしい。じれったい。

しんくくさい：
心がはればれしないようす

心気	18308	  
心気症	4869	  
心気一転	3477	  
心気を	1049	  
心気録心喜録	845	
...


心気くさい	39	30.0%
心気臭い	91	70.0%
vs.

辛気臭い	15881	70.1%
辛気くさい	6777	29.9%

Top 3 辛気 ngrams: 
辛気臭い	15855	  
辛気	10890	   <--- this includes the 6777 below
辛気くさい	6777

Strongly recommend splitting this entry. First, [notes] have limited effect for many end-users, unfortunately, in correcting kanji forms. But more importantly, I don't think 辛気 actually overlaps [1], or even vise-versa, except as orthographic mistakes.  Compare, for example, the ngrams for 辛気臭(くさ)い (25,000+)  and 心気臭(くさ)い(~100).　I think this would be better handled, if at all, as [sk] entries("spelling errors"). 

[2] appears to me to be predominantly as 辛気くさい, so I think the xref is appropriate.

@@ -23,0 +24 @@
+<xref type="see" seq="1720140">辛気臭い</xref>

gg5

@@ -26,0 +27 @@
+<gloss>irritation</gloss>

koj, daij, etc.

many sources split completely.  shinmeikai calls sense 2 kansai-ben

@@ -15 +15 @@
-<pos>&adj-na;</pos>
+<s_inf>esp. 心気</s_inf>
@@ -23,0 +24 @@
+<s_inf>esp. 辛気</s_inf>