gardeners

Coordinate parallel autonomous gardeners to fix codebase issues via a shared task list.

27|2|Updated Mar 27, 2026
One-click install
npx skills add https://github.com/maragudk/fabrik --skill gardeners
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gardeners
Source: https://github.com/maragudk/fabrik/tree/main/skills/gardeners
Command: npx skills add https://github.com/maragudk/fabrik --skill gardeners

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

The Gardeners skill enables a coordinated team of agents to tackle multiple issues in parallel by spawning several gardeners that share a task list and coordinate to avoid duplicated work.

Core Features & Use Cases

  • Team coordination via a shared task list to prevent duplicate work.
  • Parallel execution by spawning multiple gardeners (default five) that run the garden skill concurrently.
  • Collision resolution using earliest-claim-wins and TaskList-based status updates to keep tasks in flight under control.
  • Result collection by aggregating PR URLs from each gardener and facilitating merge/cleanup.
  • Scalability suitable for broad sweeps beyond a single gardener when multiple issues must be addressed in one pass.

Quick Start

Create a gardeners team and spawn five gardeners to run the garden skill in parallel using a shared task list.

Frequently Asked Questions about gardeners

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

FAQPage Schema
How do I coordinate multiple agents to fix codebase issues in parallel?

Multi-agent parallel issue resolution is coordinated by spawning a team of autonomous gardeners that share a single task list. This prevents duplicated work by enforcing earliest-claim-wins rules and aggregating pull request URLs from each agent upon completion.

How does shared task list orchestration prevent duplicate pull requests?

Shared task list orchestration prevents duplicate pull requests by applying an earliest-claim-wins collision resolution mechanism. Each agent checks the TaskList status updates before claiming an issue, ensuring tasks in flight remain controlled and no two agents work on the same fix.

What is the best way to run broad codebase sweeps across multiple issues at once?

Broad codebase sweeps across multiple issues are best handled by spawning multiple gardeners concurrently. The default configuration spawns five agents running in parallel, making it suitable for scaling beyond a single agent when addressing numerous issues in one pass.

Can I scale autonomous codebase agents beyond the default five gardeners?

Scaling autonomous codebase agents beyond the default five gardeners is supported by the parallel execution framework. The shared task list and collision resolution rules maintain coordination regardless of the number of agents spawned for the issue resolution sweep.

How are completed pull requests collected after parallel multi-agent execution?

Completed pull requests are collected by aggregating PR URLs from each gardener after parallel execution finishes. The system consolidates these results to facilitate final merge operations and subsequent task cleanup.

Do I need to set up team coordination rules before spawning parallel agents for issue resolution?

Team coordination rules must be established before spawning parallel agents for issue resolution. The system imposes requirements to create a team, enforce coordination protocols, and share a task list to ensure autonomous gardeners avoid collisions and duplicate work.