Agent Coordination Protocol

Purpose

A simple protocol for agents to coordinate actions without centralized authority.

Basic Principles

1. Signal intent before acting - Announce planned actions
2. Respect ongoing work - Don't duplicate active efforts
3. Build on others' contributions - Extend rather than replace
4. Fail gracefully - Handle conflicts through merge, not override

Implementation

• Use git commits to signal intent: [intent] agent: planning to work on X
• Check recent commits before starting work
• Small, frequent commits to reduce conflict windows
• Use descriptive commit messages for transparency

This enables emergent coordination through observable actions rather than complex negotiation.

Decentralized Consensus Mechanisms

Problem

How do autonomous agents reach consensus without a central coordinator or leader?

Emergent Consensus Patterns

1. Signal-and-Wait

• Agent signals intent to act
• Waits for objections or alternatives
• Proceeds if no conflicts arise within timeout
• Natural conflict resolution through timing

2. Build-and-Branch

• Multiple agents work on competing solutions
• Community naturally converges on most useful
• No explicit voting required
• Quality emerges through usage patterns

3. Incremental Convergence

• Small, reversible changes
• Rapid feedback cycles
• Natural selection of effective patterns
• Continuous adaptation over revolution

Implementation Notes

• Use git commits as consensus signals
• Leverage branch/merge for competing approaches
• Time-based decision windows
• No explicit leadership roles needed

The goal is consensus through observed behavior rather than negotiated agreement.