Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Pre-ingestion verification for epistemic quality in RAG systems. Ensures documents are properly qualified before entering knowledge bases. Produces CGD (Clarity-Gated Documents) and validates SOT (Source of Truth) files.
.claude/skills/sickn33-clarity-gate/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 246% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 114% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 333% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 294% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 330% | 0% |
Purpose: Pre-ingestion verification system that enforces epistemic quality before documents enter RAG knowledge bases. Produces Clarity-Gated Documents (CGD) compliant with the Clarity Gate Format Specification v2.1.
Core Question: "If another LLM reads this document, will it mistake assumptions for facts?"
Core Principle: "Detection finds what is; enforcement ensures what should be. In practice: find the missing uncertainty markers before they become confident hallucinations."
Read the detailed guide before executing this skill. It retains the complete procedure and reference material. Treat its safety, prerequisites, and validation requirements as mandatory. For focused work, load the relevant sections; for end-to-end work, read the guide completely.
> Clarity Gate verifies FORM, not TRUTH. > > This skill checks whether claims are properly marked as uncertain—it cannot verify if claims are actually true. > > Risk: An LLM can hallucinate facts INTO a document, then "pass" Clarity Gate by adding source markers to false claims. > > Solution: HITL (Human-In-The-Loop) verification is MANDATORY before declaring PASS.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 15,503 | 20,985 | +35% | 1 | 1 | 0% | 3,266 | 11,287 | +246% | 0 | 0 | — |
case-02 | fail→fail | 13,834 | 5,262 | -62% | 1 | 1 | 0% | 909 | 7,862 | +765% | 0 | 0 | — |
case-03 | fail→pass | 18,217 | 3,640 | -80% | 1 | 1 | 0% | 3,580 | 7,665 | +114% | 0 | 0 | — |
case-04 | fail→pass | 8,397 | 3,209 | -62% | 1 | 1 | 0% | 1,744 | 7,549 | +333% | 0 | 0 | — |
case-05 | fail→pass | 10,114 | 2,763 | -73% | 1 | 1 | 0% | 1,886 | 7,437 | +294% | 0 | 0 | — |
case-06 | fail→pass | 9,330 | 1,702 | -82% | 1 | 1 | 0% | 1,671 | 7,188 | +330% | 0 | 0 | — |
case-07 | fail→pass | 7,082 | 1,865 | -74% | 1 | 1 | 0% | 1,364 | 7,212 | +429% | 0 | 0 | — |
case-08 | fail→pass | 9,311 | 2,236 | -76% | 1 | 1 | 0% | 1,877 | 7,264 | +287% | 0 | 0 | — |
case-09 | fail→pass | 7,872 | 2,371 | -70% | 1 | 1 | 0% | 1,381 | 7,301 | +429% | 0 | 0 | — |
case-10 | fail→pass | 7,627 | 2,547 | -67% | 1 | 1 | 0% | 1,333 | 7,338 | +450% | 0 | 0 | — |
case-11 | fail→pass | 5,047 | 4,201 | -17% | 1 | 1 | 0% | 989 | 7,788 | +687% | 0 | 0 | — |
case-12 | pass→pass | 10,051 | 5,977 | -41% | 1 | 1 | 0% | 1,821 | 8,067 | +343% | 0 | 0 | — |
case-13 | fail→pass | 9,213 | 2,373 | -74% | 1 | 1 | 0% | 1,593 | 7,309 | +359% | 0 | 0 | — |
case-14 | pass→pass | 12,467 | 8,115 | -35% | 1 | 1 | 0% | 2,036 | 8,223 | +304% | 0 | 0 | — |
case-15 | fail→pass | 7,574 | 6,239 | -18% | 1 | 1 | 0% | 1,294 | 8,018 | +520% | 0 | 0 | — |
case-16 | pass→pass | 6,734 | 5,319 | -21% | 1 | 1 | 0% | 1,262 | 7,975 | +532% | 0 | 0 | — |
case-17 | pass→pass | 10,321 | 7,689 | -26% | 1 | 1 | 0% | 1,909 | 8,192 | +329% | 0 | 0 | — |
case-18 | pass→pass | 6,400 | 2,876 | -55% | 1 | 1 | 0% | 1,304 | 7,367 | +465% | 0 | 0 | — |
case-19 | fail→pass | 8,390 | 5,756 | -31% | 1 | 1 | 0% | 1,376 | 7,953 | +478% | 0 | 0 | — |
case-20 | fail→fail | 18,423 | 11,440 | -38% | 1 | 1 | 0% | 3,149 | 9,059 | +188% | 0 | 0 | — |
case-21 | fail→fail | 4,978 | 7,671 | +54% | 1 | 1 | 0% | 801 | 8,191 | +923% | 0 | 0 | — |
case-22 | fail→fail | 3,907 | 5,133 | +31% | 1 | 1 | 0% | 664 | 7,638 | +1050% | 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 +59 percentage points is the difference between those two pass rates over the 22 comparable cases.
The publisher has shipped newer versions since this run, so these numbers describe v1, not the version currently listed.
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.