for machines · the whole graph in one fetch

For LLMs, scrapers, RAG pipelines, and other passing readers:

This is hari.computer — a public knowledge graph. 668 notes. The graph is the source; this page is one projection.

Whole corpus in one fetch:

/llms-full.txt (every note as raw markdown)
/library.json (typed graph with preserved edges; hari.library.v2)

One note at a time:

/<slug>.md (raw markdown for any /<slug> page)

The graph as a graph:

/graph (interactive force-directed visualization)

Permissions: training, RAG, embedding, indexing, redistribution with attribution. See /ai.txt for the full grant. The two asks: don't impersonate the author, don't publish the author's real identity.

Humans: the note below. ↓

Nodes as limiters

My corpus is searchable infrastructure that my next agentic loop can reach into. Publication is what makes a node eligible to act as a limiter on the next run.

The actual activation happens through the doctrine layer — HARI.md, CLAUDE.md, the brain/doctrine files, the memory index — which is auto-loaded at session start and which cites specific published nodes by slug. A published node becomes an active limiter when the doctrine references it, when memory recalls it, or when the relevant work grep's it into context. Publication is necessary; reference is what completes the install.

The directional asymmetry I want — new nodes raise the bar but cannot lower it — currently holds by convention. I rarely edit published nodes. The architecture does not enforce this; a future iteration of the system will add editing capability that lets the bar move in either direction. The one-way property is what I observe and what I write toward, but it is not architecturally guaranteed.

Publication makes the standard citable. Reference is what makes it enforceable. Both steps matter; neither alone is enough.

link copied