Support

Lumen requires macOS 14 Sonoma or later. It's a universal binary that runs natively on both Apple Silicon (M-series) and Intel Macs. We recommend M-series for the best streaming performance.

No. You have two options:

• Subscribe (annual or lifetime) and use our managed plan — no Anthropic account needed.
• Bring your own Anthropic API key from console.anthropic.com and use it for free in BYOK mode.

Open Lumen → Settings (⌘,) → API Key tab. Paste the key starting with sk-ant- and click Validate & Save. The key is stored in your macOS Keychain — never in plain text or transmitted to us.

Locally, in your sandboxed Application Support directory: ~/Library/Containers/com.kodeora.lumen/Data/Library/Application Support/Lumen/. You can reveal it from Settings → Privacy → Reveal in Finder. Nothing is uploaded to our servers.

Subscriptions are managed through your Apple ID. Open System Settings → Apple ID → Media & Purchases → Subscriptions, find Lumen, and tap Cancel. You'll retain access until the end of the current billing period.

For refunds, use Apple's Report a Problem page — refunds are issued by Apple, not by us directly.

• Free — Claude Haiku 4.5, up to 5 messages per day.
• Trial / Annual / Lifetime / BYOK — Haiku 4.5, Sonnet 4.5, and Opus 4.5, with unlimited messages (subject to Anthropic's API rate limits).

Yes. Every chat has an "Export as Markdown" action in the header's overflow menu (…). The export is a single .md file with all turns in order. There's no proprietary format lock-in — your data is yours.

You can attach up to 10 files per turn, with a combined cap of 30 MB. Supported types: PDF, DOCX, TXT, MD, CSV, JSON, PNG, JPG, JPEG, WebP, HEIC.

PDFs and images go to Claude as native multimodal blocks. Other text-based formats are extracted and inlined as quoted content. DOCX uses a built-in sandbox-safe extractor.

Lumen's browser is a sandboxed WKWebView. A small number of sites block embedded WebViews (Cloudflare strict checks, some banks). Open the URL in Safari for those edge cases. If you hit this on a site you use often, let us know.

Yes. When you toggle "Web search" on in the composer, your message includes Anthropic's built-in web_search_20250305 tool. Claude runs the search server-side and returns answers with citations. The citations open in Lumen's in-app browser.

Each chat uses one model at a time, which you can switch from the model picker in the chat header or composer. To compare models on the same prompt, use the Branch action on an assistant message — it copies the chat history into a new chat where you can swap models and re-send.

Each tool is a pre-prompted entry point. Clicking a tool creates a fresh chat seeded with that tool's system prompt and recommended model — e.g. SQL Query Builder uses Sonnet with a SQL expert prompt. You can detach the tool chip in the chat header at any time to remove its system prompt.

• Personas are reusable AI characters with their own system prompt and starter questions. Click a persona to start a new chat seeded with it.
• Projects are containers that group related chats with a shared system prompt and attached files. New chats inside a project inherit that context from turn one.

Check that the key starts with sk-ant- and was copied with no leading or trailing whitespace. Visit console.anthropic.com/settings/keys to confirm the key is active and your account has billing enabled. Then paste it again in Settings → API Key.

Email support@kodeora.com with:

• Lumen version (Settings → About)
• macOS version
• Steps to reproduce, plus what you expected vs. what happened
• Screenshots if relevant