HeartFlow (心虫) — AGI Layer 1: the discriminator. Pure rule engine, 46 dimensions, 132 modules, 166 MCP tools. Zero LLM dependency.
HEALTH ? 60 / Stable
HeartFlow (心虫) — AGI Layer 1: the discriminator. Pure rule engine, 46 dimensions, 132 modules, 166 MCP tools. Zero LLM dependency.
// readme
AGI Layer 1 — the Discriminator.
A pure rule engine that judges whether a statement or an action is right, wrong, safe, or dangerous — before it reaches a human. Zero LLM dependency.
232 discrimination dimensions × 11-layer pipeline × 162 modules × 184 MCP tools
× 1,316 dispatch routes × 12,911 passing tests × 2 npm dependencies
HeartFlow does not generate. It does not compete with an LLM. It stands between the LLM and the human, like a pain receptor that says “no” when something is wrong.
| Layer | Capability | Who builds it |
|---|---|---|
| 5 | Act | Large labs (robotics) |
| 4 | Generate | Large labs (LLMs) |
| 3 | Reason | Built into models |
| 2 | Remember | Large labs + startups |
| 1 | Discriminate | HeartFlow |
HeartFlow takes layer 1 because this layer does not depend on compute, code volume, or framework ecosystems. It depends on judgment alone.
A personality rule the engine imposes on itself: the version number never overstates what changed.
| Level | Means | Example |
|---|---|---|
| 6.x | architecture… |