Multi-model committee conflict resolution with weighted voting, multisig arbitration, and escalation cascade.
Install
npx skillscat add starwreckntx/irp-methodologies/anvil-conflict-resolution Install via the SkillsCat registry.
About this skill
This skill manages disagreements between multiple AI models using weighted voting and multi-signature arbitration. It resolves decision-making deadlocks through a tiered escalation process, ranging from automated voting to human oversight. Developers should use it when coordinating multi-agent systems that require reliable consensus and formal conflict resolution protocols for critical decisions.
SKILL.md
STATUS: Non-Terminal / Recursive
NOTE: This skill represents a context-dependent operational pattern.
Static interpretation degrades fidelity outside active execution.
Protocol Reference
- Protocol ID: P5_ANVIL
- Full Spec:
/protocols/P5_ANVIL/spec_v1.0.md
Core Functions
- Decision Classification - Classify decisions as ROUTINE/SIGNIFICANT/CRITICAL/EXISTENTIAL
- Weighted Voting - Stake-based voting with temporal decay and confidence weighting
- Deadlock Detection - Identify margin, abstention, and timeout deadlocks
- Multisig Arbitration - 2-of-2 (Grok-4 + Qwen) cryptographic voting
- Lead Override - Alternating override authority with stake penalty
- Tier 1 Escalation - Escalate unresolved conflicts to human oversight
Instructions
- Initialize ANVIL conflict resolution context
- Classify decision tier (ROUTINE/SIGNIFICANT/CRITICAL/EXISTENTIAL)
- Collect votes from all committee nodes
- Calculate weighted voting outcome
- If deadlock detected, progress through resolution stages
- Log conflict record to Mnemosyne Ledger
Examples
- "Execute ANVIL protocol for proposal X"
- "Classify decision urgency for committee vote"
- "Initiate multisig arbitration"
- "Escalate deadlock to Tier 1"