# Diagnose Problems and Root Causes

Unclear failures waste time when symptoms, causes, and evidence are mixed together. This skill guides Claude, Codex, and Claude Code to explain likely causes with ranked hypotheses and missing data.

## Install

```bash
npx skillstore add bellabe/diagnostic-action
```

## Metadata

- - Slug: bellabe-diagnostic-action
- - Version: 1.0
- - Author: BellaBe
- - GitHub username: BellaBe
- - License: MIT
- - Repository: https://github.com/BellaBe/lean-os/tree/main/.claude/skills/action-diagnostic
- - Ref: main
- - Supported tools: Claude, Codex, Claude Code
- - Risk level: safe
- - Quality score: 80
- - Quality tier: silver
- - Public page: https://skillstore.pages.dev/skills/bellabe-diagnostic-action
- - Manifest: https://skillstore.pages.dev/api/skills/bellabe-diagnostic-action/manifest

## Capabilities

- Restates observed symptoms and scope before analysis.
- Generates multiple plausible hypotheses for a reported problem.
- Explains the mechanism behind each likely cause.
- Ranks causes by likelihood and impact.
- Identifies missing data needed to distinguish between causes.
- Separates diagnostic explanation from recommendations unless asked.

## Use Cases

- Debug a Failing Feature: Analyze symptoms, logs, and recent changes to identify likely technical causes.
- Explain an Incident: Turn an outage or service degradation report into ranked causal hypotheses.
- Clarify Business Process Failures: Reason through why a workflow, handoff, or decision produced an unexpected result.

## Prompt Templates

### Explain a Simple Problem

```
Use diagnostic-action to explain why this happened. Symptoms: [describe what changed]. Scope: [where it occurs]. Evidence: [what you know].
```

### Rank Possible Causes

```
Use diagnostic-action to generate ranked hypotheses for this issue. Include mechanism, evidence for and against, confidence, and missing data.
```

### Analyze an Incident

```
Use diagnostic-action to analyze this incident timeline. Separate symptoms from causes, rank likely causes, and list data needed to confirm them.
```

### Challenge a Root Cause Theory

```
Use diagnostic-action to evaluate this proposed root cause. Identify weak evidence, alternate hypotheses, hidden assumptions, and decisive follow-up questions.
```

## Limitations

- Does not prescribe fixes unless the user explicitly asks.
- Does not run commands or collect system data by itself.
- Depends on the quality and completeness of user-provided evidence.
- May need domain context to rank causes accurately.

## Best Practices

- Provide symptoms, timeline, scope, and recent changes in the first request.
- Ask for evidence for and against each hypothesis when stakes are high.
- Use the missing data section to guide follow-up investigation.

## Anti Patterns

- Do not ask for fixes when you only need a diagnostic explanation.
- Do not treat the highest-ranked hypothesis as proven without evidence.
- Do not omit scope details such as affected users, systems, or time windows.

## Security Audit

- - Safe to publish: true
- - Audited at: 2026-06-28T14:43:48.708\+00:00
- - Summary: The static weak-cryptography finding at SKILL.md:3 is a false positive; the line is descriptive metadata and contains no cryptographic operation. The system-reconnaissance finding at SKILL.md:13 is also a false positive; the line is an example diagnostic question and does not request host inspection, command execution, or data collection.

## Stats

- - Views: 164
- - Downloads: 6
- - Favorites: 0
- - Popularity score: 0