Strategist · Designer · Operator
A project brain an AI reads before building and updates after — 30 endpoints, self-describing.
Win: The API root teaches an agent how to use it, not just what exists.
A curated system registry with per-node memory: what the pieces are, what calls what, and why a decision was made. An AI session reads it before building and writes back after, so the reasoning survives the context window it was formed in.
A 30-endpoint API whose root is self-describing — it ships its own quick start, data model, workflow and rules alongside the endpoint list, so an agent learns the whole surface in one request instead of being handed documentation a human has to paste. 19 of those endpoints are also exposed as MCP tools, so an agent reaches the same surface either way. Per-project bearer tokens, mintable from the CLI. The hard part was not storage; it was making the thing cheap enough to keep current, because a system map nobody updates is worse than none at all.
Live at layerview.dev and in use across my own projects, including as the map for this portfolio site. Pre-launch for everyone else, waitlist open.