reposync-worker

Enforce the RepoSync GitHub protocol for agent work items, handoffs, and outcome reports.

Updated Mar 22, 2026
One-click install
npx skills add https://github.com/TECH-HY/SKILLS --skill reposync-worker-tech-hy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: reposync-worker
Source: https://github.com/TECH-HY/SKILLS/tree/main/skills/reposync-worker
Command: npx skills add https://github.com/TECH-HY/SKILLS --skill reposync-worker-tech-hy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Coordinating multiple AI coding agents on the same repository often leads to lost context, overwritten work, and untraceable decisions. This Skill enforces the RepoSync protocol so every agent reads project instructions, works through a tracked Work Item, records decisions, and leaves auditable evidence on GitHub. ## Core Features & Use Cases - Structured Work Execution: Reads AGENTS.md, .reposync/config.yaml, and advisor recommendations, then classifies the task by category, complexity, risk, and affected modules before starting. - Agent Mismatch Warnings: Compares the current agent against advisor recommendations and notifies the user when another agent is materially preferred, without transferring work automatically. - Handoffs and Outcome Reports: Provides templates for chronicles, handoffs, and outcome reports so responsibility changes and completed work remain evidence-linked to commits, PRs, and tests. - Use Case: A team runs Codex, Claude Code, and Cursor on one repository. When an agent picks up a task, it follows the Work Item workflow, logs decisions in chronicle.md, and leaves an outcome report the RepoSync Advisor can later review. ## Quick Start Open a repository containing AGENTS.md or a .reposync directory and ask the agent to follow the RepoSync protocol for the current task.

Frequently Asked Questions about reposync-worker

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

FAQPage Schema
How do I coordinate multiple AI coding agents on one GitHub repository?

Use the RepoSync protocol: each agent reads AGENTS.md and .reposync recommendations, works through a Work Item on an assigned branch, records decisions in a chronicle, and leaves an outcome report. GitHub serves as the synchronization and audit layer without a central orchestrator.

How do I hand off a coding task from one AI agent to another?

Stop at a stable checkpoint, commit or identify uncommitted work, and complete handoff.md with completed work, remaining tasks, decisions, blockers, tests, and files to review. Wait for explicit user approval before another agent takes ownership.

Which AI agents are compatible with the RepoSync Worker skill?

The skill works with Codex, Claude Code, Kimi Code, Zcode, Cursor, GitHub Copilot, ChatGPT, and generic agents. It includes an agents/openai.yaml interface file declaring support for chatgpt, codex, api, and atlas products.

Does RepoSync automatically transfer tasks between agents?

No. RepoSync never launches, commands, or reassigns agents automatically. The advisor recommends, the current agent warns about mismatches, and the user decides; every transfer requires explicit user approval.

What files must exist in a repository to use RepoSync?

The repository should contain AGENTS.md as the canonical instruction file or a .reposync directory with config.yaml, advisory recommendations, and work-items folders. The skill provides templates when the project lacks its own.