PARA Methodology
PARA is an organizational system created by Tiago Forte. It sorts all information into four categories based on actionability rather than topic. LoomBrain uses PARA as the primary structure for your knowledge graph.
The four categories
Section titled “The four categories”| Category | Definition |
|---|---|
| Projects | Time-bound efforts with a defined goal and deadline |
| Areas | Ongoing responsibilities with no end date |
| Resources | Reference material relevant to future work |
| Archive | Completed, paused, or no longer relevant items |
Projects are the most actionable. When you capture content and it relates to something you are actively building or delivering, it belongs to a Project.
Areas represent the domains you maintain over time — engineering practices, a product area, a team, a discipline. Content that informs ongoing work without belonging to a specific deliverable goes here.
Resources are reference nodes: documentation, research papers, articles, frameworks, and other material you may return to across multiple Projects and Areas.
Archive is for anything that was once active but is no longer. Archiving preserves the knowledge without cluttering active views.
How LoomBrain applies PARA
Section titled “How LoomBrain applies PARA”When content is captured, AI examines the page title, body, and any note you provided via --why. It then assigns the content to one or more matching PARA items (Projects, Areas, or Resources) from your existing graph.
Nodes can belong to multiple PARA items simultaneously. A blog post on rate limiting might be tagged to both an “API Reliability” Area and an active “Auth v2” Project.
Auto-classification
Section titled “Auto-classification”Auto-classification runs as part of the ingestion pipeline after extraction. The AI considers:
- The content of the captured page
- The
--whynote if provided - Your existing Projects and Areas (to match by relevance)
If no existing PARA item matches, the content defaults to Resources.
Manual override
Section titled “Manual override”Auto-classification is a starting point, not a verdict. You can correct or refine tags at any time:
- CLI:
lb retag <node-id>— interactive prompt to reassign PARA tags - Dashboard: Open a node and edit its tags directly
- MCP: Call the
lb_update_node_tagstool from Claude
Changes take effect immediately and do not require reprocessing.