multi-agent-coordination

Synchronize work between Antigravity and Claude Code agents

⭐2stars🔀0forks📁View on GitHub🕐Updated Jan 11, 2026⬇️1downloads

When & Why to Use This Skill

The Multi-Agent Coordination skill provides a robust framework for synchronizing workflows between Antigravity and Claude Code agents. It ensures seamless context transfer through standardized handoff protocols, task queue management, and session logging, effectively preventing operational conflicts and work duplication in complex multi-agent environments.

Use Cases

  • Agent Handoffs: Seamlessly transition project context, current status, and 'next steps' when switching between different AI agents to maintain development momentum.
  • Collaborative Task Management: Coordinate multiple agents working on the same repository by managing a shared, prioritized task queue to ensure efficient resource allocation.
  • Conflict Resolution: Prevent agents from overwriting each other's work using git-based status checks and a 'one owner' protocol for coordination files.
  • Session Observability: Maintain a persistent audit trail of significant decisions and actions taken by various agents for human review and debugging.
namemulti-agent-coordination
descriptionSynchronize work between Antigravity and Claude Code agents

Multi-Agent Coordination Skill

Use this skill to keep work synchronized between Antigravity and Claude Code.

When to Use

  • Starting a new session (check current handoff state)
  • Completing a task (update status and next steps)
  • Switching agents (write handoff note)
  • Logging significant decisions or actions

Coordination File

  • Template (public): multi_agent_coord.template.md
  • Working file (local): docs/MULTI_AGENT_COORD.md

The working file is gitignored. It contains sensitive task context that should not be public.

Session Start Protocol

  1. Read the working coordination file
  2. Check git status for any conflicts
  3. Note the current handoff state
  4. If another agent was active, acknowledge the handoff

Handoff Protocol

When ending a session or switching agents, update the working file:

## Current Handoff State

**Last Updated**: [YYYY-MM-DD HH:MM TZ]
**Active Agent**: [Your agent name]
**Status**: [What you just completed]
**Next Steps**: [What the next agent should do]
**Blockers**: [Any issues or "None"]

Task Queue Management

Add tasks with priority:

| Priority | Task | Assigned To | Status |
|----------|------|-------------|--------|
| 1 | [High priority task] | Claude Code | in_progress |
| 2 | [Medium priority task] | Antigravity | pending |

Session Logging

Log significant actions:

| Date | Agent | Action | Outcome |
|------|-------|--------|---------|
| 2026-01-10 | Claude Code | Refactored auth module | Complete |

Conflict Resolution

If git status shows changes from both agents:

  1. Do NOT overwrite the other agent's work
  2. Run git stash to save your changes
  3. Report the conflict to Rob
  4. Wait for resolution before continuing

One Owner Rule

Only ONE agent should edit the coordination file per session. If you see recent edits from the other agent, read but don't write until handoff is complete.