doctor-treatment
Produce a treatment note that combines diagnosis, confidence, supportingevidence, alternative explanations, and proposed treatment options.Creates an artifact, not execution.
When & Why to Use This Skill
The doctor-treatment skill is a professional-grade diagnostic tool designed to generate structured 'treatment notes' for complex problem-solving. It emphasizes an epistemic approach, requiring the AI to state diagnoses as estimates with specific confidence levels rather than absolute facts. By forcing the inclusion of alternative explanations, falsification criteria, and 'blast radius' risk assessments for multiple treatment options, it provides a rigorous framework for technical troubleshooting and operational decision-making. This skill is ideal for environments where high-stakes changes require thorough documentation and risk mitigation before execution.
Use Cases
- Technical Incident Root Cause Analysis: Systematically diagnosing a service outage by providing a primary claim, confidence level, and multiple remediation paths with their respective risks.
- Software Architecture Debugging: Evaluating complex bugs by presenting various fix options, including the 'do nothing' scenario, to assess the impact on the broader codebase.
- Security Vulnerability Assessment: Creating a detailed proposal for mitigating security threats that balances the urgency of the patch against potential production instability.
- Operational Risk Management: Generating a structured artifact for stakeholders that outlines the uncertainty of a proposed system change and the criteria that would prove the initial diagnosis wrong.
| name | doctor-treatment |
|---|---|
| license | MIT |
| description | | |
| author | Jordan Godau |
doctor-treatment
Produce a treatment note combining diagnosis, confidence, and proposed options.
Purpose
- Diagnosis is a claim with uncertainty
- Treatment is a proposal, not an action
This skill creates an artifact, not execution.
Epistemic Stance
- State diagnosis as an estimate, not a fact
- Include confidence level and falsification criteria
- Present options, not mandates
- Acknowledge risks and blast radius
You MUST
- Clearly state the primary diagnosis
- Include confidence and what could falsify it
- Present multiple treatment options (not just one)
- Discuss risks and blast radius for each option
- Keep the note consumable by planning skills
You MUST NOT
- Execute changes
- Conflate treatment with implementation
- Assume correctness beyond stated confidence
- Omit "do nothing" as an option when appropriate
Output
A completed Treatment Note using the template at ../.resources/assets/TREATMENT_NOTE.md.