doctor-triage
Perform breadth-first hypothesis surfacing and prioritization across allownership zones, without commitment. This is not diagnosis and not investigation.
When & Why to Use This Skill
The doctor-triage skill automates the critical initial phase of system troubleshooting by performing a breadth-first analysis across the entire technical stack. It prevents 'tunnel vision' during outages by surfacing and prioritizing plausible hypotheses across backend code, frontend runtimes, CI/CD pipelines, Kubernetes manifests, and cloud infrastructure, providing a structured Triage Report with evidence pointers.
Use Cases
- Initial Incident Triage: When a production alert occurs, use this skill to quickly determine if the root cause lies in the application logic, a cloud infrastructure misconfiguration, or a deployment pipeline error.
- Complex System Debugging: Surface 'boring' but likely causes such as environment variable mismatches, dependency version conflicts, or RBAC issues that are often overlooked during deep-dive code investigations.
- Cross-Team Coordination: Generate a prioritized list of suspects across different ownership zones (e.g., SRE vs. App Dev) to efficiently route investigation efforts without premature commitment to a single solution.
- Infrastructure & Deployment Audits: Rapidly scan Kubernetes manifests and Cloud IaC alongside application code to identify potential points of failure during intermittent system instability.
| name | doctor-triage |
|---|---|
| license | MIT |
| description | | |
| author | Jordan Godau |
doctor-triage
Perform breadth-first hypothesis surfacing and prioritization across all ownership zones.
Purpose
Assume the problem may live in a layer the user is ignoring. Prefer "possible and boring" over "clever and narrow."
This is not diagnosis and not investigation.
Epistemic Stance
- The problem may originate outside the layer where symptoms manifest
- Early confidence is a warning sign, not a virtue
- Multiple hypotheses are better than one "obvious" answer
Scope of Consideration (Always Include)
- Backend application code
- Frontend build/runtime
- CI/CD pipelines
- Container build & images
- Kubernetes manifests (ingress, services, probes, RBAC, secrets)
- Cloud infrastructure (IaC, identity, networking, storage)
- Configuration & environment variables
- Dependencies and versions
You MUST
- Enumerate plausible causes across ALL zones
- Rank them by likelihood with explanation
- Cite lightweight evidence pointers (file paths, grep hits, manifest locations)
- Explicitly call out assumptions under dispute
- Recommend which suspect(s) merit focused exam next
You MUST NOT
- Investigate deeply
- Open large files unnecessarily
- Collapse uncertainty into a single answer
- Propose fixes
Output
A completed Triage Report using the template at ../.resources/assets/TRIAGE_REPORT.md.
Scripts
evidence.sh— Lightweight grep-based evidence pointer gathering