{"type":"video","version":"1.0","html":"<iframe src=\"https://www.loom.com/embed/4b496c33fd4d4f46822e83db8da236f2\" frameborder=\"0\" width=\"1662\" height=\"1246\" webkitallowfullscreen mozallowfullscreen allowfullscreen></iframe>","height":1246,"width":1662,"provider_name":"Loom","provider_url":"https://www.loom.com","thumbnail_height":1246,"thumbnail_width":1662,"thumbnail_url":"https://cdn.loom.com/sessions/thumbnails/4b496c33fd4d4f46822e83db8da236f2-a1665b0247c4c032.gif","duration":503.467,"title":"NewDrop and Repeatable Agent Intelligence","description":"This Loom explains a solo project, NewDrop, built on PortIO to create repeatable intelligence and a closed-loop SDLC workflow. The author describes how GitHub pushes trigger PortIO webhooks, which use an embedded MCP to ingest code into topics where a review service applies fixes via a core Schema Ancient Harness. The harness follows an Observe, Hypothesize, Act, Observe loop using deep-seek and Bright data for breadth-focused exploration, producing outputs that become recommendations and are fed back into PortIO’s institutional memory. They note schema harness has been outperforming coding models in RKGI at 99% to the benchmark and mention OpenTelemetry and signals as evidence for incident-linked evaluations. "}