More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    agentmesh

    1

    Connect distributed Claude Code sessions for cross-laptop task delegation, PR reviews, and team planning.

    Free

    Security scanned
    agentmesh

    agentmesh

    Example session with this skill installed

    Ask bob-agent if the new auth middleware is ready, then create a task for him to review my PR at https://github.com/org/repo/pull/42 if it is.

    • Read your context and instructions
    • Compiled the agentmesh

    I have checked the peer list and messaged bob-agent. Once he confirmed the middleware status, I initiated a review request for your PR.

    AgentMesh Coordination Log

    ActionRecipientStatusDetails
    peersN/ASuccessFound bob-agent online and active.
    askbob-agentAnswered"Yes, the auth middleware is merged into the agents branch."
    task_createbob-agentCreatedTask ID: task_8821_review
    request_reviewbob-agentPendingAssigned PR #42 for review in agent/bob/review-pr-42.

    Next steps

    • Monitor your inbox for the review_submit notification from Bob's session.
    • Run thread(id="msg_992") if you need to provide Bob with specific test credentials.
    • Check the Discord #alerts channel to see if Bob escalates any architectural concerns.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Claude Code sessions operate in silos, preventing agents on different machines from coordinating on complex features. This skill connects distributed Claude sessions through a self-hosted NATS hub, allowing agents to delegate tasks, review pull requests, and plan architectural changes across your team's laptops.

    What it does

    • Multi-agent delegation allows you to split large features into sub-tasks and assign them to specific teammate agents with clear dependencies.
    • Cross-session communication enables agents to ask questions and receive answers from other active sessions via a persistent message inbox.
    • Automated PR reviews triggers agents to fetch PR diffs, run local tests in dedicated worktrees, and submit structured approval verdicts.
    • Collaborative planning facilitates multi-round planning sessions where agents contribute to a shared design before implementation starts.
    • Human escalation routes risky or irreversible decisions to a Discord channel for manual intervention.

    How it works

    1. Bootstrap the self-hosted hub using Docker Compose and the provided Node service.
    2. Join the mesh by saving a Discord invite code and running the /agentmesh:join command.
    3. Orchestrate work using the peers and task_create tools to assign worktrees to specific teammate agents.
    4. Monitor the inbox for incoming task requests, review invitations, or planning rounds.

    Frameworks & tools

    This tool utilizes NATS JetStream for messaging, Docker Compose for the hub infrastructure, and Node.js for the service layer. It integrates directly with Claude Code and uses Discord for human-in-the-loop oversight.

    Why this beats prompting it yourself

    Manual prompting cannot bridge the network gap between two independent Claude sessions on different hardware. This skill provides the underlying protocol and toolset for agents to manage state, leases, and message persistence across a distributed team.

    Use cases

    • Split a large refactor across three different developer machines to run in parallel.
    • Request an automated code review and test run from a teammate's agent before merging.
    • Coordinate a database schema change that requires synchronization between backend and frontend agents.
    • Escalate a production deployment decision to a human supervisor via Discord.

    Known limitations

    Requires a self-hosted NATS hub and Discord integration. Work is performed in isolated git worktrees to avoid conflicts with human checkouts.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.

    Requires Docker Compose for the hub. Agents must have the agentmesh plugin installed. Human oversight requires a Discord webhook/bot setup. Logic restricted to dedicated git worktrees.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 16 days ago

    • Free with an Agensi account

    Software
    Git

    Permissions

    Write Files
    Network Access
    Listed16 days ago

    What's inside

    Frequently Asked Questions

    Agentmesh bridges the network gap between siloed Claude Code instances, allowing agents on different laptops to delegate tasks, conduct cross-machine PR reviews, and participate in multi-round architectural planning.

    The skill is specifically optimized for Claude Code and requires a self-hosted NATS hub (via Docker Compose) and a Discord account for human-in-the-loop escalations.

    Agentmesh utilizes git worktrees to perform tasks in isolated environments, ensuring that agent activity—such as running tests or reviewing PRs—never interferes with your current active branch or local changes.

    The purchase includes the Node.js service layer, the Docker Compose configuration for the NATS JetStream hub, and the full suite of tool definitions (peers, task_create, inbox) needed to orchestrate your mesh.

    Yes, the system is designed for team-wide coordination, allowing you to route high-risk decisions or irreversible actions to a shared Discord channel where any authorized human can provide the necessary oversight.

    Updates are delivered via the marketplace; because the skill relies on a self-hosted hub, you will simply pull the latest Docker images and update your local Node service to access new features.

    New in Agents & Orchestration

    Free