# Route Ambiguous Prompts by Structure

Ambiguous prompts often blend facts, emotions, action requests, and verification needs. This skill routes the prompt by structure before the assistant answers.

## Install

```bash
npx skillstore add syntagmanull/structure-judgment
```

## Metadata

- Status: approved
- Slug: syntagmanull-structure-judgment
- Skillstore revision: r1
- Version status: missing
- Tree hash: 7e5545e3b91c07075ef3bfb194058a967a5c542fa3898d1bce9d8c835f05f89e
- Author: SyntagmaNull
- GitHub username: SyntagmaNull
- License: MIT
- Repository: https://github.com/SyntagmaNull/judgment-hygiene-stack/tree/main/structure\_judgment
- Ref: 36e07d5e13068e5be64447e8f20b427cf2cbd21a
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/syntagmanull-structure-judgment
- Manifest: https://skillstore.pages.dev/api/skills/syntagmanull-structure-judgment/manifest

## Capabilities

- Identifies whether a prompt is mainly about fact, interpretation, evaluation, action, state, evidence conflict, or verification.
- Detects premise-smuggling, escalation drift, validation capture, action masking, and text-anchoring bias.
- Routes unstable or source-sensitive claims toward verification before judgment.
- Applies safety triage when self-harm or immediate danger language appears.
- Shapes answer order so evidence is stabilized before recommendations.
- Coordinates with judgment\_hygiene and verification\_hygiene style workflows.

## Use Cases

- Stabilize Workplace Escalation Questions: Separate facts, interpretations, and proposed actions before drafting a reply, report, or escalation plan.
- Handle Mixed Emotional Prompts: Preserve distress while preventing unsupported conclusions from controlling the response.
- Route Current or Source-Sensitive Claims: Detect when current facts, laws, documents, or product details require verification before analysis.

## Prompt Templates

### Classify the Prompt Layer

```
Review this user request and identify the primary structural layer before answering: [paste request]. Explain which part must be handled first.
```

### Separate Fact and Interpretation

```
Use structure-judgment on this message: [paste message]. Separate directly supported facts from interpretations, evaluations, state, and actions.
```

### Check Verification Need

```
Analyze this request for verification needs: [paste request]. Decide whether external checking should happen before judgment or advice.
```

### Build a Routing Plan

```
Apply structure-judgment to this complex case: [paste case]. Identify primary and secondary layers, main hazard, verification trigger, and downstream order.
```

## Limitations

- It does not perform external fact checking by itself.
- It does not decide the final answer or final recommendation.
- It depends on the assistant following the routing decision consistently.
- It is not a replacement for emergency support or professional advice.

## Best Practices

- Use the routing decision to change answer order, not just to name categories.
- Stabilize facts and evidence conflicts before giving action recommendations.
- Trigger verification when current, external, or source-sensitive facts control the answer.

## Anti Patterns

- Do not validate loaded user wording as established fact.
- Do not answer escalation requests before checking the premises.
- Do not expose routing labels unless they help the user avoid a wrong conclusion.

## Security Audit

- Audited at: 2026-07-07T07:45:14.14\+00:00
- Summary: The static external-command and reconnaissance findings are false positives caused by Markdown code spans and ordinary routing prose, not executable behavior. A semantic issue remains because the skill asserts a controlled-trial approval status inside untrusted content, which can bias review or installation decisions.

## Stats

- Views: 112
- Downloads: 11
- Favorites: 0
- Popularity score: 0