Agent-Body ❤中文 · English

A plugin system treated as an organism: organs, nerve impulses, a heartbeat, reflex arcs that fire with zero model calls, sleep-time memory consolidation, closed-loop self-healing.
This page is a replay of the vitals from a real development machine — the numbers were measured, and the animation replays the real event stream. Sampled 2026-09-23 19:40.

43 organs332/332 capabilities claimed 19/19 reflex arcsarchitecture integrity 6/6

One-line install

dsh plugin --profile web add \ https://github.com/1420079678-ctrl/agent-body/releases/download/v0.1.1/dsh-external-dsh-organism-0.1.1.tgz \ https://github.com/1420079678-ctrl/agent-body/releases/download/v0.1.1/dsh-external-dsh-cortex-0.1.1.tgz \ https://github.com/1420079678-ctrl/agent-body/releases/download/v0.1.1/dsh-external-dsh-zero-residence-0.1.0.tgz

dsh plugin forwards its arguments to pnpm add inside the profile, so one command installs three organs · github.com/1420079678-ctrl/agent-body

Vitals

Body · organs by call volume (every one the vitals have a record for)

The number in each cell is that organ's cumulative call count, read from vitals.json. Hover for successes / failures and fatigue.

Proprioception · the real pulse stream

Real records from pulse.jsonl, replayed on a loop: 💓 heartbeat · 🫁 pulmonary circulation · ✓ organ call · 🧩 organ change. Each heartbeat packages the current directives, the vitals and any homeostasis warnings into a blood packet and pumps it system-wide. Pulse lines are verbatim tool output and stay in the language the machine ran in — that is what makes them evidence rather than decoration.

Token gating (reproducible scope)

48 representative commands · cold start · tool-schema tokens only
Ungated: 55,154 tok on average
Gated: 8,433 tok on average
84.71% of tool-schema tokens gated away (74.25% on a live install that has run history)

The benchmark, the frozen corpus and the CI baseline all live in the repository; npm run bench:check fails the build when the number drifts.

The healing ledger and self-training