Getting Started

Create your account

Sign up at app.arclume.dev using GitHub OAuth or email and password. GitHub OAuth is recommended — it streamlines repo connections later.

Create an organization

After signing in, create an organization. All maps, repos, and team members are scoped to your organization. You can invite team members as Owner, Lead, or Member.

Connect a GitHub repository

1. Navigate to your organization's Repos page
2. Click Connect Repository
3. Install the ArcLume GitHub App on your account or organization
4. Select the repositories you want to connect
5. ArcLume will begin indexing your codebase automatically

Indexing parses your code, creates vector embeddings, and detects interface points (REST endpoints, message queues, and more). This typically takes 1–5 minutes depending on repository size.

Create a map

Maps are the central workspace in ArcLume. Each map groups related stories and transcripts. Create one per feature area or initiative.

Generate your first stories

1. Open your map and write a project outline in the Outline tab — describe your goals, target users, and key features
2. Optionally add meeting transcripts or briefs in the Transcripts tab
3. Go to the Stories tab and click Generate with AI, or ask the Chat panel to "Generate stories from my outline"
4. Stories appear in real time as Claude creates them, grounded in your indexed codebase

Each generated story includes a title, description, acceptance criteria, complexity estimate (S / M / L), and implementation context referencing your actual code.

Export to your tools

Once you're happy with the generated plan:

• Generate a context bundle — from the Outline tab, click Context Bundle and choose your target system (Lovable, Cursor, Claude Code, Windsurf, or Generic). The bundle combines your outline with relevant code snippets for pasting into your AI coding assistant. See Context Bundles.
• Push to Linear — export stories directly to your Linear project board. See Linear Integration.
• Use the MCP server — connect ArcLume to your IDE so your AI assistant has full context of your project plan and codebase structure. See MCP Server.

Next steps

• Learn about maps and stories
• Understand how codebase indexing works
• Visualize your codebase with the explore graph
• Chat with your codebase