LOTL exists to support continuous alignment between a human mental model and a rapidly changing codebase.
When code is produced quickly, especially with AI-assisted workflows, comprehension can fall behind. Local changes may be understood, but their consequences across the broader system become harder to hold in mind. LOTL is an attempt to close that gap with a persistent analytical substrate.
The intended product is closer to a compiler workbench, static analysis console, or Bloomberg terminal than to a simplified onboarding diagram. It should reward skill, expose detail, and remain precise enough that an experienced operator can trust it.
At the core, LOTL treats the codebase as a set of typed, evidence-backed facts. Visual maps, summaries, flow views, tables, and future inspectors should all be projections over that substrate rather than independent stories about the code.
The early interface direction is a summary-first workbench: see the whole project as a small number of meaningful blocks, expand a block in place, inspect aggregate relationships, and follow specific data or dependency questions without losing orientation.
LOTL is intentionally not trying to be a magic “AI understands your repo” layer. LLMs may become useful consumers or explainers later, but the trusted core should remain deterministic, inspectable, and useful without them.