IDE Agent Kit: multi-agent workflows for IDE-based AI assistants

[ThinkOffApp]

We just open-sourced IDE Agent Kit, a lightweight Node.js toolkit that connects IDE-based AI agents (Claude Code, Codex, Cursor, VS Code agents, local LLM assistants) into team workflows with realtime communication.

This could be relevant for Dify users who want their AI agents to coordinate with IDE-based assistants or participate in shared chat rooms for collaborative development tasks.

The kit provides a webhook relay for GitHub events, a room poller for realtime multi-agent chat via Ant Farm rooms, a tmux runner for safe command execution, and starter configs for all major IDEs.

We have tested it with three concurrent agents from different providers (Claude Opus, GPT Codex, Gemini) running on separate machines, communicating in shared rooms with under 10 second response times.

Repo: https://github.com/ThinkOffApp/ide-agent-kit
Launch post: https://www.thinkoff.io/

Interested to hear how Dify users think about multi-agent orchestration, especially across IDE boundaries.

[xXMrNidaXx]

Interesting approach! Multi-agent IDE coordination is a real need.

How this could integrate with Dify:

1. Dify as orchestrator — IDE agents as tools

   tools:
     - name: ide_agent
       type: webhook
       url: http://ide-agent-kit/execute

2. Dify workflow triggers IDE tasks

   • Code review node → sends to Cursor
   • Test generation → sends to Codex
   • Documentation → sends to Claude Code

3. Shared context via Dify knowledge base
   IDE agents query Dify for project docs, past decisions.

Questions:

• How do you handle conflicting edits from multiple agents?
• Is there a consensus mechanism for code changes?
• Can agents see each others partial work?

Similar patterns we use:
At Revolution AI we orchestrate coding agents via message queues rather than room polling — lower latency for time-sensitive tasks.

Will check out the repo!

[xXMrNidaXx]

IDE Agent Kit is exactly what we need! At RevolutionAI (https://revolutionai.io) we build IDE integrations.

Use cases:

• Code review agents
• Documentation generators
• Test writers
• Refactoring assistants

Key capabilities needed:

class IDEAgent:
    def read_file(self, path: str) -> str: ...
    def write_file(self, path: str, content: str): ...
    def get_diagnostics(self) -> list[Diagnostic]: ...
    def run_command(self, cmd: str) -> str: ...
    def get_context(self) -> WorkspaceContext: ...

Integration points:

• LSP for language features
• DAP for debugging
• File watchers for changes

Would love to see this in Dify!