Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Figure out whether 'the ick' about someone you're dating is a real incompatibility, a genuine red flag, or an anxious/avoidant self-sabotage pattern worth pushing through — by interrogating the specific ick honestly. Use when someone says 'I caught the ick and I don't know why', 'is this a red flag or am I just scared', 'I always find a reason to end things', or is talking themselves out of someone good. Produces a decode of the specific ick, a red-flag vs pattern verdict, and a next move. Hones
.claude/skills/mohitagw15856-the-ick-decoder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | 43% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-20 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-15 | ✓→✗ | ▼ Worse | 182% | 0% |
| case-17 | ✓→✗ | ▼ Worse | 187% | 0% |
"The ick" — the sudden inexplicable repulsion toward someone you were into — is one of dating's great confusions, because it comes in three completely different flavors wearing the same costume. Sometimes it's your instincts correctly flagging a real incompatibility or red flag. Sometimes it's an anxious/avoidant self-sabotage pattern that manufactures a reason to flee anyone who gets close. And sometimes it's a superficial ick (he ran for the bus funny) that has nothing to do with anything. Most dating advice validates whichever one you want to hear. This skill instead interrogates the specific ick to tell which flavor it is — because dumping someone great over an avoidant spiral and ignoring a real red flag are opposite mistakes with the same feeling.
pointing at (values/incompatibility, safety/red-flag, attachment-pattern, or genuinely superficial)
villain) · self-sabotage pattern (worth examining before acting) · superficial (let it pass) — with the reasoning, not a rubber stamp
history (the tell of self-sabotage)
with it a beat, or notice-and-let-go
Ask for (if not already provided):
three days running", not "something felt off")
when it's getting real? do they ignore red flags for too long? (both are patterns)
undecodable; the specific trigger is everything. Get concrete — the exact behavior, moment, or trait — because "he was too available" and "he lied about where he was" are wildly different data wearing the same discomfort.
that predicts harm — control, dishonesty, cruelty, boundary-pushing. Trust it. (b) Real incompatibility: a genuine values/life/attraction mismatch — valid, and nobody's the villain. (c) Attachment pattern: the ick reliably appears right as intimacy deepens, especially with people who are actually kind and available — the signature of avoidant self-sabotage. (d) Superficial: a texture/quirk thing that your brain inflated — usually passes.
does this ick fit a cycle? If the user consistently gets the ick from stable, kind, into-them partners right when it's getting real, the ick is more about the user's nervous system than the person. If they consistently override real red flags until it blows up, that's the opposite pattern. History decodes the present.
avoidant patterns often manufacture icks precisely about a partner's warmth, availability, and effort (the exact good qualities) — because closeness itself is the threat. A red flag is about behavior that would hurt you. Same feeling, opposite meaning; name which one this is.
user what they want to hear. Red flag → here's how to end it cleanly and safely. Incompatibility → it's okay to leave, no cruelty needed. Pattern → before you bolt, here's the experiment (name it to yourself, sit a week, see if it's the person or the closing distance). Superficial → let it go, or don't, but know that's what it is.
## Your specific ick
[The exact trigger, stated concretely]
## What flavor is it?
[Red flag / real incompatibility / attachment pattern / superficial — with the
reasoning that places it there]
## The pattern check
[Does this fit a cycle in your history? Available-people ick vs override-red-flags —
which, if either?]
## Available-ick vs unsafe-ick
[Is this about them being warm/into you, or about behavior that would hurt you?]
## Your honest next move
[End cleanly (+ how) / leave without a villain / run the pattern experiment first /
notice and let it pass — the real one, not the flattering one]self-sabotage from a real flag, which sometimes means the unwelcome answer
both lazy takes; the specific trigger decides
is a red flag, and safety always overrides "maybe it's your pattern"
"no villain" is a valid ending
deserve real therapeutic work, and leave it there
dating-profile-doctor]] for the front of the funnel; franklin-decision-ledger]] when it's a genuine should-I-stay call; future-self-interview]] for the longer view on a relationship fork.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 16,799 | 15,933 | -5% | 1 | 1 | 0% | 2,585 | 3,233 | +25% | 0 | 0 | — |
case-02 | pass→pass | 18,942 | 18,415 | -3% | 1 | 1 | 0% | 2,390 | 3,250 | +36% | 0 | 0 | — |
case-03 | pass→pass | 23,213 | 13,819 | -40% | 1 | 1 | 0% | 2,473 | 3,389 | +37% | 0 | 0 | — |
case-04 | pass→pass | 19,516 | 13,317 | -32% | 1 | 1 | 0% | 2,282 | 3,200 | +40% | 0 | 0 | — |
case-05 | fail→pass | 20,561 | 26,221 | +28% | 1 | 1 | 0% | 2,314 | 3,319 | +43% | 0 | 0 | — |
case-06 | pass→pass | 24,451 | 14,715 | -40% | 1 | 1 | 0% | 2,920 | 3,575 | +22% | 0 | 0 | — |
case-07 | fail→pass | 24,077 | 13,613 | -43% | 1 | 1 | 0% | 2,699 | 3,361 | +25% | 0 | 0 | — |
case-08 | pass→pass | 21,347 | 17,432 | -18% | 1 | 1 | 0% | 2,332 | 3,237 | +39% | 0 | 0 | — |
case-09 | pass→pass | 19,872 | 18,373 | -8% | 1 | 1 | 0% | 2,565 | 3,373 | +32% | 0 | 0 | — |
case-10 | pass→pass | 26,183 | 11,808 | -55% | 1 | 1 | 0% | 2,644 | 3,290 | +24% | 0 | 0 | — |
case-11 | pass→pass | 20,152 | 12,595 | -38% | 1 | 1 | 0% | 2,354 | 2,607 | +11% | 0 | 0 | — |
case-12 | pass→pass | 20,630 | 11,727 | -43% | 1 | 1 | 0% | 2,248 | 3,366 | +50% | 0 | 0 | — |
case-13 | pass→pass | 17,517 | 18,102 | +3% | 1 | 1 | 0% | 2,315 | 3,209 | +39% | 0 | 0 | — |
case-14 | pass→pass | 22,320 | 19,092 | -14% | 1 | 1 | 0% | 2,475 | 3,584 | +45% | 0 | 0 | — |
case-15 | pass→fail | 14,340 | 22,618 | +58% | 1 | 1 | 0% | 1,310 | 3,689 | +182% | 0 | 0 | — |
case-16 | pass→pass | 13,997 | 14,795 | +6% | 1 | 1 | 0% | 1,881 | 2,987 | +59% | 0 | 0 | — |
case-17 | pass→fail | 26,234 | 40,699 | +55% | 1 | 1 | 0% | 2,860 | 8,208 | +187% | 0 | 0 | — |
case-18 | pass→pass | 22,369 | 17,173 | -23% | 1 | 1 | 0% | 2,225 | 3,324 | +49% | 0 | 0 | — |
case-19 | pass→pass | 21,408 | 16,434 | -23% | 1 | 1 | 0% | 2,086 | 3,268 | +57% | 0 | 0 | — |
case-20 | fail→pass | 25,528 | 11,108 | -56% | 1 | 1 | 0% | 2,705 | 3,070 | +13% | 0 | 0 | — |
case-21 | pass→pass | 22,383 | 18,768 | -16% | 1 | 1 | 0% | 2,535 | 3,655 | +44% | 0 | 0 | — |
case-22 | pass→pass | 21,963 | 15,906 | -28% | 1 | 1 | 0% | 2,358 | 3,088 | +31% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.