What people build on Lyt
Use casesOne knowledge layer covers the situations below: an agent that starts a session already briefed, an existing markdown folder brought in as it stands, a decision log an agent can cite, a team sharing one versioned source, documentation you subscribe to, projects kept apart, and a machine rebuilt from Git.
Nine situations
Each one has its own page: what the situation is, how a person actually does it with their agent, and what changes as a result.
Give your agent your markdown notes
Your notes stay markdown in a vault, and your agent searches the reachable pod instead of starting every session blind.
Read this use caseBring in a markdown folder you already have
Adopting an existing Obsidian vault or plain folder is additive: your files, folder names, and links stay exactly as they were.
Read this use caseKeep a decision log an agent can cite
A captured decision is one markdown file an agent can quote back with its path, rather than paraphrasing from memory.
Read this use caseReturn to what you concluded months ago
Ranked search across the pod brings an old conclusion back with its wording, its date, and the file it lives in.
Read this use caseKeep one team knowledge base in Git
Shared vaults live in Git with the same review and history model as code, and access is granted explicitly, per vault.
Read this use caseBring someone new onto the team's material
Receiving a shared vault puts a local copy on the new colleague's machine and indexes it, so their own agent answers from day one.
Read this use caseFollow documentation that keeps its origin
Subscribe to a public documentation repository and get a local read-only copy your agent reads, refreshed when you sync.
Read this use caseWork across several projects without mixing them
Meshes are named groups of vaults — scoping a search to one mesh keeps a client's question inside that client's material.
Read this use caseReconstruct your workspace on another machine
Because the vaults are Git repositories, a replaced machine is a rebuild from them, with the local indexes made again locally.
Read this use caseWhat they have in common
In every case the markdown stays readable and locally available under its access mode — owned vaults are yours to edit, subscribed copies are yours to read — and the agent is the one operating it. Each page sets out the situation, then gives the ask you can copy into your own agent, with the command it runs underneath for anyone who prefers to type it. The mechanism is the same throughout — vaults of markdown in Git, local indexes rebuilt per machine, and explicit boundaries at the edge.
Start with one vault
However a case above begins — with your own notes or with a vault someone else keeps — the place to start is your own first vault: one vault, one note, one search.