Vvelox Labs

VeloxLabs (“Vellox”) is an AI tooling and frameworks lab attached to the San Francisco advisory practice Velox Intelligence, focused on Information Architecture, AI Legibility, and human-AI Semantic Interaction Architecture.

This repository is the early public workspace for Vvelox's tooling, consulting aides, and public web legibility systems.

Projects

• Vvelox Labs Skills Suite — Open-source and proprietary tools and consulting aides with portable Claude skill folders and explicit human/agent safety boundaries.
  • Task Breakdown Skill — The first skill, goblin-task-breakdown, turns ambiguous task language into a stable executable task graph. Compatible with the folder contract used by my colleague, Moses Liao's claude-skills; see the bridge note.
• AI Framework Schema — A human-first interaction framework and structured context schema for LLMs, autonomous agents, and agentic workflows (runtime embodiment).
• Human-in-the-Loop Protocols — State-gated context continuity and human-in-the-loop protocols for iterative and recursive agentic pipelines (runtime embodiment).
• AI Lblr — A transparency label and badging model for AI-legible public web systems.
• Sweet Suite — A public web root-file kit for AI Legibility, including Sweet Web Root Files such as robots.txt, sitemap.xml, llms.txt, schema.jsonld, and mermaid.js.

Focus areas

• AI Legibility
• AI Visibility
• Semantic Interaction Architecture
• JSON-LD
• Public web infrastructure
• Transparency labels
• Human-AI workflow governance
• Executable task graphs and consulting aides

Maintainer

Tess McCarthy · vveloxltd.com