A mesh is a declarative configuration that defines how multiple AI agents interact, route messages, and manage state within a specific workflow or application.

Can I pass dynamic variables between agents?

Yes, the Mesh Builder supports template interpolation using {key} tokens that resolve from message payloads or workspace variables at runtime.

What is the benefit of using a dispatcher for routing?

The dispatcher mode centralizes routing logic in your YAML config, allowing agents to stay focused on tasks while the system handles targets, branching, and fan-out logic.

How do I handle complex agent pipelines?

The skill recommends the Phase Coordinators pattern, which breaks large workflows into discrete, single-responsibility coordinators to reduce context accumulation and errors.

What should I avoid writing in my agent prompts?

Do not write message protocols, routing instructions, rearmatter formats, or tool usage details, as these are automatically injected by the TX system.

TX Mesh Builder

Name: TX Mesh Builder
Author: eighteyes

byeighteyes

•

生産性とワークフロー

Architects complex multi-agent workflows and declarative agent configurations for the TX V4 harness.

The Mesh Builder skill empowers developers to design, test, and deploy sophisticated agentic meshes using the TX declarative framework. It provides specific guiderails for defining agent roles, managing state-aware routing, and implementing advanced patterns like Phase Coordinators and Fan-out/Fan-in dispatching. By automating the injection of message protocols and workspace context, it allows creators to focus purely on agent logic and domain-specific decision trees rather than boilerplate infrastructure.

主な機能

01Standardized agent boundaries to prevent behavior bleed in small-model coordinators

029 GitHub stars

03Automated protocol injection for message frontmatter, rearmatter, and workspace structure

04Declarative multi-agent routing including linear, branching, and parallel fan-out patterns

05Dynamic prompt template interpolation with automatic workspace path resolution

06Phase Coordinator implementation to manage complex pipelines with single-responsibility agents

ユースケース

01Creating state-persistent narrative engines that resolve dynamic paths for per-session game data

02Building a sequential content pipeline with specialized research, drafting, and editing agents

03Developing a parallel code review system using fan-out routing and automated haiku summarization

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add eighteyes/tx mesh-builder

For use in Claude.ai and ChatGPT