OpenHands/agent-canvas

Refactor dev scripts: rename dev-safe and add modular service launch options

已關閉

#226 建立於 2026年5月9日

 (7 則留言) (0 個反應) (0 位負責人)TypeScript (26 個分叉)github user discovery
enhancementgood first issue

倉庫指標

星標
 (43 顆星)
PR 合併指標
 (平均合併 2天 10小時) (30 天內合併 106 個 PR)

描述

Problem

The current dev script naming is confusing:

  • dev:safe isn't intuitive - it's really "agent-server + canvas only"
  • dev:minimal is an alias for dev:safe adding to confusion
  • No clear naming convention for different service combinations

Current Scripts

Script What it runs
dev All services (agent-server + automation + canvas)
dev:automation All services (same as dev)
dev:safe / dev:minimal Agent-server + Canvas
dev:frontend Canvas only (assumes running backend)
dev:extra-backend Extra standalone agent-server

Proposed Solution

Rename scripts with a clear naming convention based on which services are included:

New Name Old Name Services
dev dev:automation All (default: agent-server + automation + canvas)
dev:canvas - Canvas only (assumes running backend)
dev:server dev:safe Agent-server + Canvas
dev:server:only - Agent-server only (no canvas)
dev:full dev:automation Explicit all services
dev:extra-backend - Extra standalone agent-server (keep as is)

CLI Options (Nice-to-have)

The underlying script could support flags to selectively enable/disable services:

node scripts/dev.mjs --no-canvas        # Agent-server + automation only
node scripts/dev.mjs --no-automation    # Agent-server + canvas only  
node scripts/dev.mjs --server-only      # Just agent-server

Implementation Notes

  1. Rename dev-safe.mjs to something clearer (e.g., dev-server.mjs)
  2. Consolidate logic into a unified dev.mjs script with service toggles
  3. Keep backward compatibility aliases during transition
  4. Update DEVELOPMENT.md and AGENTS.md with new naming

This issue was created by an AI agent (OpenHands) on behalf of the user.

貢獻者指南