distributed-mesh

Configures git-based state synchronization for squads distributed across multiple machines.

1|Updated Jul 7, 2026
One-click install
npx skills add https://github.com/seiggy/maf-copilot-studio-demo --skill distributed-mesh-seiggy
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: distributed-mesh
Source: https://github.com/seiggy/maf-copilot-studio-demo/tree/main/.squad/templates/skills/distributed-mesh
Command: npx skills add https://github.com/seiggy/maf-copilot-studio-demo --skill distributed-mesh-seiggy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jq, git, curl.

What problem does it solve? When agent squads run on different machines, CI runners, or partner organizations, their state files are not on the local filesystem. This Skill sets up a git-based transport layer so remote squad state materializes locally before agents read it, without daemons, servers, or federation protocols. ## Core Features & Use Cases - Mesh topology configuration: Generates a mesh.json file mapping each squad to a zone (local, remote-trusted, or remote-opaque) with the appropriate connection details. - Bundled sync scripts: Copies pre-built sync-mesh.sh and sync-mesh.ps1 scripts that pull Zone 2 git repos and fetch Zone 3 published contracts via HTTP. - State repo initialization: Scaffolds a shared mesh state repository with per-squad directories and placeholder SUMMARY.md files using the --init mode. - Use Case: A developer laptop squad and a CI squad need to share results. Configure mesh.json with the CI squad as remote-trusted, run sync-mesh.sh before agent startup, and agents read the pulled state as ordinary local files. ## Quick Start Ask the agent to set up a distributed mesh for your squads and answer its questions about which squads participate and which zone each belongs to.

Frequently Asked Questions about distributed-mesh

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

FAQPage Schema
How do I share state between agent squads on different machines?

Configure a mesh.json file listing each squad with its zone, then run sync-mesh.sh or sync-mesh.ps1 before agents start. Remote-trusted squads sync via git clone or pull, and remote-opaque squads are fetched as published SUMMARY.md contracts over HTTPS.

What is the difference between remote-trusted and remote-opaque zones?

Remote-trusted squads are in the same organization with shared git access, so their full state is pulled from a git repository. Remote-opaque squads are external partners with no shared auth, so you only fetch the SUMMARY.md contract they publish over HTTPS.

Does the distributed mesh require a running server or daemon?

No. The mesh uses plain git push and pull plus curl fetches, executed on demand before agents start and after they finish. There are no daemons, message queues, service discovery, or persistent processes involved.

How are write conflicts avoided in the shared mesh state repo?

Each squad writes only to its own directory, such as its own board, squad folder, and dated drop files. Because no two squads write to the same file, git push and pull never conflict, and a failed push is fixed with git pull --rebase followed by push.

When should I not use the distributed mesh pattern?

Skip it when all squads are on the same machine, since agents can read local files directly with zero transport. It is also unnecessary if you expect real-time updates, because the pattern delivers recent-enough state via periodic git pulls, not live notifications.