LUMI review
A free, open-source VS Code agent extension with approval-gated edits and no inline completion.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
LUMI is an open-source AI coding assistant extension for developers who want an agent to work with project context inside VS Code. It reads workspace files and offers Plan and Act modes for planning changes, editing files, and running terminal commands. Its approval gates keep mutating actions under developer control. The product is also listed for Cursor, and its supported platforms include Windows, macOS, and Linux. It is aimed at solo and small-team use, particularly for developers comfortable directing an agent through chat rather than relying on inline code completion.
Context handling is a notable strength: chats can include files, folders, problems, terminal output, git diffs, and URLs, and LUMI supports MCP server connections. Checkpoints allow rollback after tool execution, while local task history and cognitive memory support continuity across work. Governed parallel subagent execution adds a way to split work across agents. These features suit repository-level tasks where the developer wants to provide relevant material and review proposed actions. Developers looking primarily for inline suggestions as they type should choose an extension with that capability instead.
LUMI is free and open source, with model connections including OpenRouter, ChatGPT Subscription, NousResearch, Cloudflare Workers AI, Cerebras, ClinePass, Grok/X Subscription, Qwen Token Plan, and Z AI. That range gives developers options for connecting models, alongside MCP servers, but the agent workflow still depends on selecting and connecting a provider. The published product details do not establish a paid-plan comparison; the listed support channel is community. LUMI is a good fit for solo developers who value configurable, approval-based agent work and can rely on community support. Teams that need inline completion or a formal support channel should look elsewhere.
LUMI pros and cons
- Where it wins
- Reads workspace files and accepts rich context such as diffs and terminal output
- Approval-gated file edits and terminal commands, with checkpoints for rollback
- Connects multiple model providers and supports governed parallel subagents
- Where it doesn't
- No inline code completion
- Community support only
- Agent workflow requires approval for mutating actions
LUMI fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. How we rank.
Last updated · How we research and update