amElnagdy/delegate-skills

Delegate a coding task to a separate coding agent CLI, review the diff, land the commit yourself — one per implementer.

What it solves

delegate-skills provides a standardized way to orchestrate multiple AI coding assistants (implementers) from a single orchestrating agent. It solves the problem of fragmented toolsets by allowing users to organize different AI CLIs into "fleet lanes" (e.g., feature, tests, UI) and delegate specific tasks to the most appropriate tool for that job, while ensuring the human reviewer retains control over the final commit.

How it works

The project uses a "relay" system to bridge an orchestrating agent and various implementer CLIs (like Aider, Claude Code, Cursor, and others).

  1. Fleet Setup: The delegate-setup skill discovers installed CLIs on the machine and proposes a configuration of "lanes" that bind specific types of work to specific implementers.
  2. Dispatch: When a task is delegated, the orchestrator writes a brief and dispatches it via a relay script (relay.mjs).
  3. Execution: The relay launches the implementer CLI, manages the run, and captures the output into a structured result.json file.
  4. Review Loop: The implementer edits the working tree, but the relay never commits the changes. The user or orchestrator reviews the diff and runs project gates before manually landing the commit.

Who it’s for

Developers who use multiple AI coding tools and want a unified interface to manage them through an orchestrating agent without sacrificing the review-and-commit workflow.

Highlights

  • Broad Implementer Support: Includes relays for a wide array of AI CLIs including Aider, Claude Code, Cursor, Cline, Grok, and others.
  • Standardized Contract: Every relay follows the same delegate-relay.result.v1 contract, making implementers interchangeable.
  • Review-First Workflow: Strictly enforces that the relay never commits changes, keeping the commit authority with the human reviewer.
  • Zero-Dependency Relays: Relays are built using only Node.js built-ins to ensure security and transparency.
  • Fleet Management: Allows for global or project-specific configuration of lanes to map work types to specific AI tools.

Related

  • Project
    awslabs/cli-agent-orchestrator

    A coordination framework that allows a supervisor AI to delegate work to multiple specialist AI coding CLIs in parallel or sequence.

  • Project
    codeaholicguy/ai-devkit

    A control plane for AI coding agents that provides a unified configuration, a supervision console, and local searchable memory to coordinate multiple AI assistants.

  • Project
    KKKKhazix/khazix-skills

    A collection of standardized AI agent skills (instruction sets) for goal definition, disk cleanup, deep research, and project documentation maintenance.

  • Project
    dyoshikawa/rulesync

    A Node.js CLI tool that synchronizes AI development rules and configurations across multiple AI coding tools from a single unified source.

  • Project
    zhinkgit/embeddedskills

    A set of skills for AI coding assistants to autonomously control compilers, debuggers, and communication buses for a complete embedded AI development loop.