Devil Chain
Multi-perspective serial review with convergence loop. Each perspective corrects the blind spots of the last.
Effect
Extends checklist-driven Devil into a serial chain. Multiple specialist perspectives run in sequence; if any fix is made, the entire chain reruns.
Security → Performance → Logic → Flow
↓ Concern fixed → rerun chain
↓ All perspectives clear → convergedWhen to Use
- Changes span 3+ files or cross multiple layers (API + DB + UI, etc.)
- A single-perspective Devil feels insufficient to catch everything
- Final check before shipping to production
How It Works
Step 1: Security
Hardcoded API keys / secrets in logs / input validation & injection / auth & authorization / information leakage on error
Step 2: Performance
N+1 queries / expensive operations inside loops / client & connection reuse / memory footprint / redundant computation
Step 3: Logic
Edge cases & off-by-one / return type consistency / Null/None handling / inconsistency with existing patterns / default value coherence
Step 4: Flow
End-to-end data flow integrity / environment variable fallback consistency / legacy compatibility & rollback feasibility / external service contracts
Loop Verdict
- Concerns found → fix → rerun the entire chain as Round N+1
- All 4 steps pass, zero concerns → declare convergence
Combinations
| Card | Synergy |
|---|---|
| Devil Gate | Load the relevant checklist at each step (precision UP) |
| Devil Starswarm | Parallelize the 4 chain steps across Agents (speed UP, cost UP) |
”…A single pair of eyes has its limits. That’s why you change the eyes.” — Eris
Want to use this card permanently? Convert it directly into a SKILL.md — Effect → Overview, When to Use → Trigger, How It Works → Steps.