cdesktop
A local browser interface for coding-agent sessions and diff review.
Claude Code in Claude Desktop serves people working on graphical coding-agent sessions, code changes and review. These independent projects explore parts of that workflow with documented AI assistance. They are candidates to evaluate, not guaranteed drop-in replacements.
Reviewed 2026-10-29. These facts describe the original tool; evaluate each alternative against your specific requirements.
A local browser interface for coding-agent sessions and diff review.
Capability statements come from project documentation. We have not functionally tested these third-party apps.
| Project | Documented scope | Cost / license | Platform | Self-hosting | AI provenance |
|---|---|---|---|---|---|
| cdesktopexperimental · Repository | free Apache-2.0 | web | Documented | Maker claim |
Switchable workflow: Manage coding-agent sessions and review code changes in a local browser interface in place of the Claude Desktop Code tab. Source
The project describes itself as beta. It runs as a local web interface; native desktop installers are not yet shipped. It wraps separately installed agents and their usage costs and permissions still apply. Project source
Compare Claude Code in Claude Desktop vs cdesktopUse the step-by-step switching checklist
Match your actual workflow against the documented scope and limitations above. Compare a coding session, file changes and review on a disposable repository. A local interface that calls external agents still depends on the selected agent and its own permissions. None of these listings is a promise of feature parity.
Each app links to attributable evidence of AI-assisted development. A verified contribution badge means a public commit credits an AI coauthor; a maker claim means the owner states AI was used. Neither establishes that every line was AI-generated or reviewed.
cdesktop are documented as free to use. Check the license linked for each project. Running infrastructure can still incur costs. Open source alone is not evidence of AI authorship.
For Claude Code in Claude Desktop, focus on coding sessions, diff review, agent and provider support. Keep originals, try disposable copies and inspect round-trip exports. Missing comparison facts are unknown.