Node Only Agent Loop

Node Only Agent Loop

This page shows a local loop that works from npm without Python.

The goal is practical: preserve useful candidate work, keep missing work visible, and produce a JSON report that another agent or CI job can inspect.

Create Demo Files

npx pic-ts demo bootstrap --output-dir .pic-demo --overwrite

This creates small JSON files for a runtime state, runtime input, packet example, and phase request.

Run A Runtime Step

npx pic-ts runtime step \
  --state .pic-demo/runtime_state.json \
  --input .pic-demo/runtime_step_input.json \
  --output .pic-demo/runtime_step_report.generated.json

The result is a JSON report. It can be useful even when settled=false.

Export And Inspect A Packet

npx pic-ts packet export \
  --report .pic-demo/runtime_step_report.generated.json \
  --output .pic-demo/packet.json

npx pic-ts packet inspect --packet .pic-demo/packet.json

Packet content is inert data. The inspector may detect command-like text, but it does not run it.

Build A Phase Plan

npx pic-ts phase plan --request .pic-demo/asi_proxy_phase_request.json --compact
npx pic-ts phase gap --request .pic-demo/asi_proxy_phase_request.json --compact

Candidate-only intake does not reduce the phase gap. Missing obligations and blockers remain visible.

Run The Compact Agent Acceleration Report

npx pic-ts agent accelerate --compact --text "Candidate packet: preserve residuals." --profile development

This is a report generator. It does not execute shell commands or mutate a repository.

What This Loop Means

The loop is useful for Node.js agent runtimes because it produces structured reports that can be reused safely.

It does not prove real ASI, physical outcomes, simulator truth, policy truth, or oracle truth. It preserves unfinished work and shows the next finite checks.