team-bernie

Coordinate parallel multi-agent workflows with mailbox-based planning and file edit locks.

2|Updated Apr 2, 2026

Search Tags:#multi-agent #orchestration #task dependencies #agent collaboration #project governance #planning approval #lock-based concurrency

Ready to run

npx skills add https://github.com/cassioalexandre/agf --skill team-bernie

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: team-bernie
Source: https://github.com/cassioalexandre/agf/tree/main/.agents/skills/Team%20Bernie
Command: npx skills add https://github.com/cassioalexandre/agf --skill team-bernie

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill solves the problem of getting multiple AI agents to collaborate safely and coherently on the same project without stepping on each other’s changes.

Core Features & Use Cases

Parallel multi-agent orchestration: Runs specialists (frontend, backend/DB, research, marketing, reviewer) concurrently while staying aligned.
Planning gatekeeping: Forces agents to submit an Action Plan to Bernie and wait for approval before making significant changes.
File edit safety with locks: Prevents simultaneous edits by using .antigravity/team/locks/ semaphores and requiring lock release after completion.

Quick Start

Use the team-bernie skill to initialize Bernie’s team orchestration for your project, then ask for parallel tasks (e.g., backend + frontend + documentation) while Bernie manages approvals and lock-safe execution.

Frequently Asked Questions about team-bernie

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I coordinate parallel AI agents to prevent conflicting file edits?▼

You can coordinate parallel AI agents by using a shared team state directory that tracks task dependencies and applies lock-based mutual exclusion. Agents must acquire file locks before editing and release them after completion to prevent conflicts.

What is mailbox-based planning in multi-agent orchestration?▼

Mailbox-based planning in multi-agent orchestration routes Action Plans from specialist agents to a central leader agent for review. The leader must approve each plan before specialists execute significant project changes.

How do I manage task dependencies across concurrent AI agents?▼

Managing task dependencies across concurrent AI agents requires a shared team state directory that records task statuses and dependency tracking. This structure ensures dependent tasks wait for upstream completion before initiating execution.

Can I run frontend, backend, and research AI agents concurrently on the same project?▼

Yes, you can run frontend, backend, and research AI agents concurrently on the same project. A leader agent broadcasts messages to align specialists, while planning approvals and file locks ensure changes are synchronized safely.

How do file locks protect multi-agent workflow execution?▼

File locks protect multi-agent workflow execution by acting as semaphores that enforce mutual exclusion. They prevent multiple agents from writing to the same file simultaneously, requiring explicit lock release after task completion.

Do I need a shared state directory to synchronize agent team execution?▼

Yes, a shared state directory is required to synchronize agent team execution. It stores task status and dependency tracking data alongside lock semaphores, which are essential for coordinating parallel work and preventing conflicts.