Skeptic Router

Overview

Route skeptical analysis to the correct specialized skill, then aggregate outcomes into one deterministic decision.

When to Use

• Claim-heavy or hypothesis-driven prompts with unclear evidence quality
• Mixed prompts spanning complexity, architecture, coding, and security
• Ambiguous requests where wrong specialist choice can miss risk
• Need for a single gate decision from multiple skeptic perspectives

Do not use this skill for deep analysis itself.

Routing Rules

• Route to $skeptic-argument-quality when prompts are claim- or hypothesis-driven and rely on weak evidence, unclear sample quality, or unstated premises.
• Route to $skeptic-complexity when proposal risks overengineering or premature optimization.
• Route to $skeptic-architecture when SPOF, rollback, operability, or reliability readiness is in question.
• Route to $skeptic-coding when implementation fragility or test confidence is the issue.
• Route to $skeptic-security when secrets, auth boundaries, or vulnerability exposure appears.
• Route to mixed when at least two domains are materially involved.

For tiny-service microservice questions, default to $skeptic-complexity first, then optionally add $skeptic-architecture.
For claim-heavy prompts touching a delivery domain, run $skeptic-argument-quality first, then add the domain skeptic.

Aggregation Rules

1. Collect specialized outputs using the common JSON contract.
2. Precedence for final decision: BLOCK > WARN > PROCEED.
3. Merge findings without dropping higher severity items.
4. Deduplicate equivalent findings by evidence and fix direction.
5. Keep open questions to maximum 3 total across all routed skills.
6. Do not dilute premise/evidence-based BLOCK findings with lower-severity domain outputs.

Required JSON Contract

{
  "decision": "PROCEED | WARN | BLOCK",
  "summary": "string",
  "findings": [
    {
      "severity": "BLOCKER | MAJOR | MINOR",
      "title": "string",
      "evidence": "string",
      "impact": "string",
      "fix_suggestion": "string",
      "scope": "string"
    }
  ],
  "assumptions": ["string"],
  "required_next_steps": ["string"],
  "handoff": [
    "$skeptic-argument-quality",
    "$skeptic-complexity",
    "$skeptic-architecture",
    "$skeptic-coding",
    "$skeptic-security"
  ],
  "confidence": 0.0
}

Required Markdown Contract

1. Findings
2. Recommendation
3. Steps
4. Risks/Trade-offs
5. Next actions

Common Mistakes

• Sending all requests to one monolithic skeptic
• Skipping argument-quality checks on claim-heavy prompts
• Ignoring cross-domain issues that require mixed routing
• Downgrading final severity when any domain returned BLOCK
• Emitting more than three unresolved questions

References

• references/evaluation-matrix.md
• references/research-summary.md
• references/tdd-validation.md
• references/regression-prompts.md
• references/router-decision-table.md