Feature parity: All Claude.ai/Desktop features should be available in Claude Code

Resolved 💬 4 comments Opened Jan 1, 2026 by wodge73 Closed Mar 5, 2026

Summary

Claude Code provides an incredibly flexible platform for power users. Being terminal-based, it's highly customisable, scriptable, and integrates seamlessly into developer workflows. However, some features available in the Claude desktop/web app are not available in Claude Code.

Power users choosing Claude Code should not have to sacrifice functionality.

The principle

  • ✅ Claude Code can have more features than desktop (e.g., MCP servers, multiple email account connections, terminal access, file system access) — this is expected for a power user tool
  • ❌ Claude Code should not have fewer features than desktop — power users shouldn't be limited

Features currently missing from Claude Code (vs Claude.ai/Desktop)

  • Research mode — extended autonomous web research (see #15982)
  • Projects — persistent context and knowledge bases
  • Artifacts — interactive previews of generated content
  • Canvas — collaborative document editing
  • File uploads with vision — drag-and-drop image/PDF analysis in conversation flow
  • Conversation branching — explore alternative responses
  • Starred conversations — organisation features

Why this matters

Claude Code's terminal-based approach offers unique advantages:

  • Scriptable and automatable
  • Composable with other CLI tools
  • Works over SSH/remote connections
  • Integrates with existing developer toolchains
  • More flexible context management (MCP, custom tools)
  • Can connect to multiple accounts/services simultaneously

Users who prefer this power-user interface shouldn't have to switch to the web UI just to access features like Research mode or Projects.

Request

Please prioritise feature parity so that all capabilities available in Claude.ai/Desktop are also available in Claude Code. The terminal interface should be a superset of features, not a subset.

View original on GitHub ↗

This issue has 4 comments on GitHub. Read the full discussion on GitHub ↗