KDCube Application Configuration: Six Data Classes, One Live Contract
How KDCube separates platform, app, and user properties and secrets, then resolves live configuration through async SDK contracts.
KDCube Writing
Engineering deep dives, industry perspective, implementation notes, and practical recipes.
Search covers titles, summaries, tags and full article text.
A local model as your agent's brain
Point a KDCube agent at a model on your own machine — served by Ollama — and let a user pick it for a single conversation. A small models gateway translates the existing provider: custom path; a reserved descriptor property turns it on.
Instruction profiles: a pickable brain-style for your agent
An agent's instructions compose from blocks, so it can offer named profiles a user picks per conversation — including a distilled set for local models.
The Self-Governing ReAct Round: State Machine, Live Gates, and Honest Feedback
A KDCube ReAct turn is a state machine that runs the model, watches what it generates as it streams, decides in real time what may reach the user, and — when something goes wrong — tells the model exactly what happened before asking it to continue.
Settle My Solution in KDCube: The Wrap, Not the Rewrite
Host your working agent — LangGraph, CrewAI, a raw loop — as a KDCube app with its framework intact. The wrap is a thin async host layer, not a rewrite.
App Ingredients: Build the Boundary Before the Features
One small, production-shaped KDCube app — surface, contracts, identity, state, tests, release. Twelve stamped operations, from app card to shipped.
Your Agent Is Not a Prototype
Your team's agent works. The executive brief on the third path to production: keep it, wrap it thin, and let the runtime do the rest.
The Second Project: Running an AI Agent for Real Users
Your agent works. The second project is the AI agent infrastructure around it — identity, spend, evidence, updates. A runtime takes it off your roadmap.
MCP Security Best Practices: The Plug Is Standard. The Permissions Are Yours.
MCP plugs your agents into the world’s tools — and your product into the world’s agents. The executive brief: who may, on whose authority, who pays.
MCP Security Architecture: Why Connecting AI Agents Is the Easy Part
MCP makes agent connectivity portable. The enterprise challenge is governing which agents receive which tools, whose authority crosses the endpoint, who pays, and what evidence remains.
Keep the Agent. Add the Runtime.
Your agent already works. Production should keep it — the framework, the behavior, the edge you built — and add the runtime around it.
Before Your AI Agent Acts: An Executive Brief on Governance
Seven plain questions that decide whether your AI agent governance is real — and what it means for each answer to be enforced, not written down.