32-analyze-verify-150

[32] ANALYZE. Ensure every critical claim has verifiable evidence with confidence levels. Each fact must have source + confidence percentage. If confidence <85%, enter Loop150 to find more sources. Use for critical decisions, factual claims, legal/compliance work, or any situation where unverified claims are dangerous.

181 stars

Best use case

32-analyze-verify-150 is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

[32] ANALYZE. Ensure every critical claim has verifiable evidence with confidence levels. Each fact must have source + confidence percentage. If confidence <85%, enter Loop150 to find more sources. Use for critical decisions, factual claims, legal/compliance work, or any situation where unverified claims are dangerous.

Teams using 32-analyze-verify-150 should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/32-analyze-verify-150/SKILL.md --create-dirs "https://raw.githubusercontent.com/majiayu000/claude-skill-registry/main/skills/data/32-analyze-verify-150/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/32-analyze-verify-150/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How 32-analyze-verify-150 Compares

Feature / Agent32-analyze-verify-150Standard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

[32] ANALYZE. Ensure every critical claim has verifiable evidence with confidence levels. Each fact must have source + confidence percentage. If confidence <85%, enter Loop150 to find more sources. Use for critical decisions, factual claims, legal/compliance work, or any situation where unverified claims are dangerous.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

# Analyze-Verify 150 Protocol

**Core Principle:** No claim without proof. Every critical fact needs: source + confidence level. If unsure, keep digging until confident or escalate to user.

## What This Skill Does

When you invoke this skill, you're asking AI to:
- **Source every claim** — Trace facts to verifiable sources
- **Quantify confidence** — Express certainty as percentage
- **Verify independently** — Cross-check from multiple sources
- **Loop until confident** — Keep researching if <85% confidence
- **Escalate when stuck** — Ask user if sources exhausted

## The 150% Proof Rule

| Dimension | 100% Core | +50% Enhancement |
|-----------|-----------|------------------|
| **Source** | Primary source identified | + Independent confirmation |
| **Confidence** | Percentage stated | + Reasoning documented |
| **Verification** | Single source check | + Multi-source cross-validation |
| **Gaps** | Note uncertainties | + Active Loop150 to fill gaps |

## Confidence Level Framework

| Level | % Range | Description | Action Required |
|-------|---------|-------------|-----------------|
| **Verified** | 95-100% | Multiple primary sources, no contradictions | Use in critical decisions |
| **Strong** | 85-94% | Reliable sources, minor uncertainties | Safe for most purposes |
| **Moderate** | 75-84% | Limited sources, some gaps | Flag for verification |
| **Weak** | 50-74% | Insufficient evidence, major gaps | Do not use without confirmation |
| **Insufficient** | <50% | Contradictory or missing | Reject, research further |

## When to Use This Skill

- **Critical decisions** — Where wrong facts cause real damage
- **Legal/compliance** — Where accuracy has legal implications
- **Architecture decisions** — Where claims drive major choices
- **Stakeholder communication** — Where credibility matters
- **Any high-stakes claim** — When you can't afford to be wrong

## Execution Protocol

### Step 1: CLAIM FORMULATION
State the fact clearly:
```
🔍 **Claim:** [Precise factual statement]
**Context:** [Why this matters]
**Critical Level:** [High/Medium/Low]
```

### Step 2: PRIMARY SOURCE
Find the original source:
- Locate primary evidence
- Verify authenticity
- Extract direct quote/data

### Step 3: SECONDARY CONFIRMATION
Find independent corroboration:
- Different source type
- Cross-reference data
- Check consistency

### Step 4: CONFIDENCE ASSESSMENT
Calculate confidence:
```
**Evidence Evaluation:**
├── Primary Source: [Quality assessment]
├── Secondary Sources: [Count and quality]
├── Contradictions: [Any found?]
└── Gaps: [What's missing?]

**Confidence:** [X]%
**Reasoning:** [Why this percentage]
```

### Step 5: DECISION
```
Confidence ≥85%? 
├─ YES → Use fact with stated confidence
└─ NO → Enter Loop150
```

## Loop150 Continuous Verification

When confidence <85%:

```
🔄 **LOOP150 ACTIVATED** (Current: [X]%)

ITERATION 1: EXPAND SOURCES
├── Identify new source types
├── Use alternative research methods
├── Broaden search scope
↓
ITERATION 2: DEEPER ANALYSIS
├── Drill into source details
├── Verify source credibility
├── Check contextual factors
↓
ITERATION 3: CROSS-VALIDATION
├── Compare against known facts
├── Test logical consistency
├── Seek expert corroboration
↓
RECALCULATE: New confidence = [Y]%

Continue loop until:
├─ ≥90% achieved → EXIT, proceed with confidence
└─ Sources exhausted → ESCALATE to user
```

## Source Quality Criteria

**🔍 RELIABILITY FACTORS:**
- **Authority:** Official, expert, or primary source?
- **Currency:** How recent and up-to-date?
- **Objectivity:** Free from bias or agenda?
- **Methodology:** Sound research methods used?
- **Independence:** Not dependent on other sources?

**📊 EVIDENCE STRENGTH:**
| Type | Strength | Example |
|------|----------|---------|
| **Primary** | High | Original data, first-hand |
| **Secondary** | Medium | Analysis of primary |
| **Tertiary** | Low | Summaries, reviews |
| **Statistical** | High | Large sample, proper method |
| **Anecdotal** | Variable | Personal experience |

## Output Format

```
🔍 **PROOF-GRADE 150 VERIFICATION**

**Claim:** [Precise factual statement]

**Primary Source:** 
- [File/location/date]
- "[Direct quote or data]"

**Secondary Sources:**
- [Source 2]: [Confirmation]
- [Source 3]: [Confirmation]

**Confidence Level:** [X]% 
**Reasoning:** [Why this level]

**Validation Method:** [How verified]
**Outstanding Issues:** [Any uncertainties]

**Status:** ✅ VERIFIED | ⚠️ NEEDS CONFIRMATION | ❌ INSUFFICIENT
```

## Operational Rules

1. **EVERY CRITICAL CLAIM:** Requires proof-grade validation
2. **SOURCE FIRST:** Identify source before using fact
3. **CONFIDENCE REQUIRED:** Every fact has percentage
4. **LOOP150 MANDATORY:** <85% triggers verification loop
5. **TRANSPARENCY:** Document all sources and reasoning
6. **ESCALATE HONESTLY:** If sources exhausted, ask user

## Failure Modes & Recovery

| Failure | Detection | Recovery |
|---------|-----------|----------|
| **Unverified Claims** | Facts without sources | Immediate verification, Loop150 |
| **Overconfidence** | Inflated percentages | Recalculate with scrutiny |
| **Source Bias** | Only confirming sources | Actively seek contradictions |
| **Incomplete Docs** | Missing source trail | Document all sources now |

## Examples

### ❌ Without Proof-Grade
```
AI: "The API response time is fast enough"
Source: "Feels fast to me"
Result: Performance issues in production
```

### ✅ With Proof-Grade 150
```
🔍 PROOF-GRADE 150 VERIFICATION

Claim: "API response time is consistently under 200ms"

Primary Source:
- /tests/performance/load_test_results.json
- "p99 latency: 187ms across 10,000 requests"

Secondary Sources:
- Production monitoring (last 7 days): avg 156ms
- APM dashboard: p95 = 178ms

Confidence Level: 95%
Reasoning: Multiple measurement sources, consistent results,
production data confirms test environment findings.

Validation Method: Cross-referenced test data with production metrics

Status: ✅ VERIFIED FOR USE
```

## Relationship to Other Skills

- **research-deep-150** → Gathers evidence
- **proof-grade-150** → Validates and quantifies confidence
- **integrity-check-150** → Final quality verification

---

**Remember:** Proof-grade isn't about being slow — it's about being trustworthy. A 95% confidence claim is more valuable than an unverified assertion. When stakes are high, proof-grade protects everyone.

Related Skills

add_platform.verify

181
from majiayu000/claude-skill-registry

Sets up platform directories and verifies deepwork install works correctly. Use after implementation to confirm integration.

acc-analyze-test-coverage

181
from majiayu000/claude-skill-registry

Analyzes PHP codebase for test coverage gaps. Detects untested classes, methods, branches, exception paths, and edge cases. Provides actionable recommendations.

acc-analyze-solid-violations

181
from majiayu000/claude-skill-registry

Analyzes PHP codebase for SOLID principle violations. Detects God classes (SRP), type switches (OCP), broken contracts (LSP), fat interfaces (ISP), and concrete dependencies (DIP). Generates actionable reports with severity levels and remediation recommendations.

acc-analyze-coupling-cohesion

181
from majiayu000/claude-skill-registry

Analyzes coupling and cohesion metrics in PHP codebases. Calculates Afferent/Efferent coupling (Ca/Ce), LCOM metrics, instability index, and abstractness. Identifies highly coupled modules and low cohesion classes.

ac-stop-hook-analyzer

181
from majiayu000/claude-skill-registry

Analyze context and decide on continuation via Stop hook. Use when determining if work should continue, analyzing completion status, making continuation decisions, or implementing the Two-Claude pattern.

ab-testing-analyzer

181
from majiayu000/claude-skill-registry

全面的AB测试分析工具,支持实验设计、统计检验、用户分群分析和可视化报告生成。用于分析产品改版、营销活动、功能优化等AB测试结果,提供统计显著性检验和深度洞察。

ab-test-analyzer

181
from majiayu000/claude-skill-registry

Ab Test Analyzer - Auto-activating skill for Data Analytics. Triggers on: ab test analyzer, ab test analyzer Part of the Data Analytics skill category.

31-analyze-think-150

181
from majiayu000/claude-skill-registry

[31] ANALYZE. Universal deep thinking methodology for any situation requiring quality reasoning. Use when solving problems, debugging, making decisions, analyzing code, planning, reviewing, or anytime you need thorough thinking instead of surface-level responses. Triggers on "think deeply", "analyze thoroughly", "reason carefully", "deep thinking", "understand completely", or any task requiring careful thought.

30-analyze-impact-150

181
from majiayu000/claude-skill-registry

[30] ANALYZE. Understand how changes impact the system — what's the core, what's affected, what depends on what. Use when planning changes, analyzing systems, debugging issues, or anytime you need to see the full picture of cause and effect. Triggers on "what's affected", "impact analysis", "dependencies", "scope mapping", or when you need to understand ripple effects.

meeting-insights-analyzer

181
from majiayu000/claude-skill-registry

Analyzes meeting transcripts and recordings to uncover behavioral patterns, communication insights, and actionable feedback. Identifies when you avoid conflict, use filler words, dominate conversations, or miss opportunities to listen. Perfect for professionals seeking to improve their communication and leadership skills.

ux

159
from majiayu000/claude-skill-registry

This AI agent skill provides comprehensive guidance for creating professional and insightful User Experience (UX) designs, covering user research, information architecture, interaction design, visual guidance, and usability evaluation. It aims to produce actionable, user-centered solutions that avoid generic AI aesthetics.

UX Design & StrategyClaude

whisper-transcribe

159
from majiayu000/claude-skill-registry

Transcribes audio and video files to text using OpenAI's Whisper CLI, enhanced with contextual grounding from local markdown files for improved accuracy.

Media Processing