/proof
// proof the method survives contact with real systems — some you can check, some is the pattern, generalized
// start here
Three links: a coding-agent plugin, a platform delivery case, and the workflow that keeps the work inspectable.
brew + npm
AgentOps
external
Coding-agent plugin for any model: Claude, GPT, and more. Go CLI + skills for memory, validation, tracked work, and reusable operating loops.
infrastructure
Dify Agent Platform
internal
A self-hosted AI application platform adapted for restricted OpenShift with observability, repeatable deployment, and operator-safe handoff.
process
RPI Workflow
internal
A delivery loop that turns AI work into research, plans, validation, implementation waves, and reusable learning.
// proof lanes
Grouped by the kind of confidence each one should create for a visitor.
// make ai work repeatable
Tools and operating principles that preserve context, force validation, and turn agent sessions into compounding work.
brew + npm
AgentOps
go / typescript / claude-code / cli
Coding-agent plugin for any model: Claude, GPT, and more. Go CLI + skills for memory, validation, tracked work, and reusable operating loops.
methodology
12-Factor AgentOps
github / docs
Operational principles for making AI agent work reviewable, testable, and safe to ship.
npm
vibe-check
typescript / cli
A validation CLI that reads git history and turns AI-assisted coding sessions into measurable reliability signals.
// self-host ai you can operate
Kubernetes and OpenShift patterns for self-hosted agent platforms, RAG systems, and restricted environments where install, observability, rollback, and recovery matter.
infrastructure
Dify Agent Platform
helm / gitops / kubernetes
A self-hosted AI application platform adapted for restricted OpenShift with observability, repeatable deployment, and operator-safe handoff.
gov
OpenShift PaaS
kubernetes / gpu / gitops
Enterprise container-platform patterns for restricted self-managed environments: multi-tenant, GPU-aware, GitOps-driven.
ops
Platform Tooling
python / bash / automation
Automation for upgrades, diagnostics, secrets, backups, recovery, and day-2 operations where reliability has to be inspectable.
// convert inputs into judgment
A visible loop for writing, practicing, and turning patterns into reusable delivery systems.
archive
Writing
next.js / mdx
Essays and devlogs documenting the operating loop, what I am seeing as the work compounds.
process
RPI Workflow
git / beads / .agents
A delivery loop that turns AI work into research, plans, validation, implementation waves, and reusable learning.
upstream
gastown-operator
go / kubernetes / operator-sdk
Kubernetes operator work exploring how agent orchestration becomes a durable control plane.
Want the operating loop behind the artifacts?
Builds are the receipts. The method is what makes them repeatable.