From skills-for-humanity
Diagnoses broken reasoning—flawed arguments, contradictory specs, unsound plans, or invalid inferences—and produces a corrected version. Use when reasoning needs fixing, not just critiquing.
How this skill is triggered — by the user, by Claude, or both
Slash command
/skills-for-humanity:s4h-logic-fixerThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Diagnosis without repair is incomplete. This skill takes broken reasoning and produces a corrected version — one where the premises actually support the conclusion, the hidden assumptions are made explicit, the fallacies are removed, and the argument can be defended.
Diagnosis without repair is incomplete. This skill takes broken reasoning and produces a corrected version — one where the premises actually support the conclusion, the hidden assumptions are made explicit, the fallacies are removed, and the argument can be defended.
The output is not a critique. It is a fixed version of what you were trying to say.
Step 1: Diagnose before repairing Don't jump to a fix. First, be specific about what's broken:
Framing check: Confirm the argument being examined before continuing. State what you've identified — the actual argument being analyzed, the conclusion it is trying to establish, and the apparent failure type — in one sentence, then use AskUserQuestion:
Step 2: Classify the repair needed Different failure modes need different repairs:
| Failure | Repair |
|---|---|
| Unsupported premise | Add evidence, or qualify the premise |
| Overclaimed conclusion | Narrow the conclusion to what the premises actually support |
| Missing step | Make the implicit inference explicit |
| Circular reasoning | Identify the begged question; provide independent support for the conclusion |
| False dichotomy | Reframe to acknowledge the full option space |
| Invalid inference | Add the bridging premise that makes the inference valid |
| Equivocation | Disambiguate the term; use different words for different senses |
| Hidden assumption | Surface it as an explicit premise; assess whether it holds |
Step 3: Produce the fixed version Rewrite the argument so that:
Step 4: Note what changed and why Don't just hand over the fixed version without explanation. For each change:
If the original conclusion cannot be made to hold under any reasonable repair — because the premises are too weak or the claim is simply unsupported — say so directly rather than producing a version that technically avoids fallacies but doesn't actually establish what the author wanted to establish.
Before proceeding, use the AskUserQuestion tool. State your interpretation of the situation in 1–2 sentences — what is being analyzed and what the core question is — then ask:
Proceed based on their selection. If the user reframes, incorporate the correction before running any analysis.
Original reasoning: [The broken argument, spec, or reasoning as provided]
Diagnosis
| Issue | Location | Type |
|---|---|---|
| [what's broken] | [where in the argument] | [failure type] |
Fixed Version [The corrected argument — written cleanly, as if it were the original. Not a tracked-changes edit, but a finished version.]
What Changed
| Original | Fixed | Why |
|---|---|---|
| [original claim/step] | [fixed version] | [reason] |
Strength Assessment [Is the fixed argument now sound? Does it establish what the original was trying to establish, or did the repairs require narrowing the conclusion? Be honest about this.]
Sometimes a broken argument can't be fixed because the conclusion isn't supportable — the premises, even repaired, don't get you there. In those cases, the honest output is: "This conclusion cannot be established with the evidence available. Here is what the available evidence does support." That is more useful than a technically valid argument for a weaker claim dressed up to look like the original.
The fixer's job is sound reasoning, not salvaging a predetermined conclusion.
After delivering this output, use AskUserQuestion to offer the next move:
/s4h-logic-check — Verify the fix holds under full scrutiny/s4h-decision-premortem-analysis — Stress-test the corrected plan/s4h-communication-clarity-audit — Check the corrected reasoning is communicated clearlynpx claudepluginhub human-avatar/skills-for-humanityProduces a complete logic report on any argument, plan, or reasoning — validates premises, tests inference, detects fallacies, and surfaces hidden assumptions.
Strips AI-slop patterns from chain-of-thought, extended thinking, and agent decomposition traces—not final prose. Targets over-explaining questions, hedging plans, trivial over-decomposition, infinite-loop rationalization.
Construct well-structured arguments using the hypothesis-argument-example triad. Use for PR descriptions, ADRs, code review feedback, and technical proposals.