Keep the reasoning around your code close to the code itself. Living Architecture Nodes shows when architecture memory is missing, stale, or orphaned before the next human or AI change compounds the drift.
Core operation stays local and does not require an account.
Architecture notes, dependency assumptions, regression context, and module intent are easy to leave behind. That becomes more expensive when another developer or an AI system continues from incomplete context.
Living Architecture Nodes treats architecture memory as a maintained part of the repository rather than a document somebody remembers to update later.
Scan significant source files and identify modules that do not yet have a companion .node.md memory file.
Flag node memory that has fallen behind its source file so the next change starts with an explicit warning instead of silent drift.
Export local JSON and Markdown handoff bundles so another human or AI tool can start from structured repository context instead of guessing.
One sidebar view for missing, stale/dirty, and orphan node memory.
Create missing .node.md files from a structured template after explicit confirmation. Drafts are not treated as verified architecture truth.
Jump directly to ARCH.md, NERVE.md, and CHANGELOG.node.md when the repository uses them.
Write diagnostic context to .lan-vscode/ as local JSON and Markdown.
Read-only scanning remains available without Workspace Trust. File generation and exports require trust.
Core Free operation does not send source code, architecture memory, diagnostics, secrets, or behavioral telemetry to a remote service.
Open a workspace in VS Code and select Living Architecture Nodes from the Activity Bar.
Review missing, stale, dirty, or orphan node memory before making a broad change.
In a trusted workspace, generate structured companion node drafts for missing areas.
Populate and verify the draft. Generated structure is a starting point, not a claim of architectural truth.
Create local diagnostic context when another developer or AI system needs to continue the work.
The extension reads the active workspace for the checks you request. It writes only when you explicitly generate node drafts or export a handoff bundle, and those write paths are confined to the trusted workspace.