Governed agent systems, applied workflows, open research, and published developer tools—with the status and evidence of each project made explicit.
A governed operating method for AI agents working across complex, multi-step workflows.
Separates context, roles, authority, process state, and verification. Agents can act within bounded responsibilities, but completion is determined by evidence—not by the model declaring itself done.
Controlled retrieval trace & visibility scanner for AI web crawlers.
Inspects what AI models receive and return when ingesting web applications. Surfaces retrieval boundaries, extractable evidence, and machine readability without guesswork.
Deterministic weather verification & insurance evidence engine.
Ingests meteorological data from traceable sources to calculate deterministic findings and evidence grades for climate claims under 60 seconds.
Open specification proposal for structured, agent-readable websites.
Built after observing that llms.txt provides text summaries but not structured agent interfaces — validated on 50,665 links across 1,711 files.
Zero-trust CI/CD pipeline and multi-service deployment discipline.
A structured delivery blueprint for multi-service TypeScript applications. Strict 5-layer dependency rules, versioned Docker builds, automated health checks, secret sync audits, and instant rollback paths.
CLI & library tool to validate web endpoints and AI manifests against index-ai spec rules.
Detects and fixes environment variable drift across microservices and CI/CD environments.
Strict runtime schema validation for application configuration.