evidence-based-analysis

This skill should be used when presenting "investigation findings", "design decisions", "code review feedback", or any analysis that requires structured evidence. Provides evidence tables, confidence levels (High/Med/Low), fact vs speculation separation, and source citation patterns.

仓库: flox/forge-plugin
最近来源活动: 2026年3月10日 18:08
检测到的 SKILL.md 语言: 英语
星标: 2
分支: 0

安装方式

默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。

检查来源文件

决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。

返回搜索返回安装

name	evidence-based-analysis
description	This skill should be used when presenting "investigation findings", "design decisions", "code review feedback", or any analysis that requires structured evidence. Provides evidence tables, confidence levels (High/Med/Low), fact vs speculation separation, and source citation patterns.

Evidence-Based Analysis

Structured patterns for gathering facts, analyzing evidence, and presenting findings in investigation and design work.

Core Principle

Facts before opinions. Evidence before actions.

Present what you observed, cite your sources, and distinguish certainty from speculation. Let evidence drive conclusions, not assumptions.

When to Use This Pattern

Always use for:

Investigation findings (issue triage, root cause analysis)
Design decisions (architectural choices, trade-offs)
Code review feedback (concrete issues with evidence)
Research summaries (external patterns, best practices)

Pattern applies when:

Presenting findings that will inform user decisions
Making claims that need verification
Documenting analysis for future reference

Evidence Table Format

Use structured tables to present findings clearly:

| Finding | Source | Confidence |
|---------|--------|------------|
| {what was found} | {file, commit, issue, doc} | High/Med/Low |
| {what was found} | {file, commit, issue, doc} | High/Med/Low |

Finding: Concise statement of what you discovered

Source: Where the evidence came from:

Code: file.py:line or file.py:line-range
Git: commit abc123 or PR #123

Level	Criteria	When to Use
High	Direct evidence (code, logs, commits, tests)	You read it yourself
Medium	Inferred from patterns (similar cases)	Based on precedent
Low	Possible based on architecture (hypothetical)	Educated guess

✅ Good examples:
- "The git log shows commit abc123 on 2026-02-15"
- "File auth.py:148 catches all exceptions without re-raising"
- "The API returns HTTP 500 for invalid tokens"

✅ Good examples (marked clearly):
- "Hypothesis: The refactor broke token validation"
- "Likely cause: The recent auth changes introduced this"
- "Assumption: Mobile clients use the same endpoint"

❌ Bad examples (unmarked speculation):
- "This is probably broken"
- "The auth module doesn't work"
- "Users will be confused"

## Summary

{2-3 sentence executive summary}

## Evidence

| Finding | Source | Confidence |
|---------|--------|------------|
| {finding 1} | {source} | High/Med/Low |
| {finding 2} | {source} | High/Med/Low |

## Analysis

{Reasoning from evidence to conclusions}

## Recommendations

1. {Highest priority action}
2. {Second priority action}
3. {Lower priority or optional action}

## Decision: {What are we deciding}

**Context:** {Why is this decision needed?}

**Options Considered:**

1. **{Option A}**
   - **Pros:** {Advantages}
   - **Cons:** {Disadvantages}
   - **Evidence:** {Prior art, constraints}

2. **{Option B}**
   - **Pros:** {Advantages}
   - **Cons:** {Disadvantages}
   - **Evidence:** {Prior art, constraints}

**Decision:** {What we chose and why}

**Consequences:** {Trade-offs we're accepting}

**Verification:** {How we'll know this was the right choice}

## Code Review Findings

### Critical Issues (Must Fix)

**Item 1: {Title}**
- **File:** {file}:{line}
- **Issue:** {What's wrong}
- **Evidence:** {Why it's wrong — error condition, contract}
- **Fix:** {Recommended solution}

### Important Issues (Should Fix)

**Item 2: {Title}**
...

### Minor Issues (Nice to Have)

**Item 3: {Title}**
...

# Pattern: file:line or file:line-range
src/auth/api.py:148
src/auth/api.py:142-156

# Commit reference
Commit abc123d (2026-02-15): "refactor: Simplify auth handling"

# PR reference
PR #123: "Add token expiration handling"

# Context files
.forge-context/context/product.md (Integration Points section)

# External docs
https://jwt.io/introduction (JWT structure and validation)

Aspect	Pattern	Purpose
Evidence table	Finding → Source → Confidence	Structured fact presentation
Confidence levels	High/Med/Low	Clear uncertainty communication
Fact vs. speculation	Explicit markers	Prevent misleading claims
Source citation	file:line, commits, docs	Verifiable references
Structured presentation	Summary → Evidence → Analysis → Recommendations	Logical flow