Tag: AI
All the articles with the tag "AI".
Selecting more than one tag in the sidebar shows only posts that have every selected tag.
- #AI#Kagent#SRE
Diagnosing OpenCTI with Kagent (6) - Drill 2: Invoice Failure After the SRE Upgrades
A blind rehearsal drill against the upgraded L402 monitoring. The pages fired in 2 minutes 21 seconds and the agent found the broken component, but it skipped a required tool and invented details. A blameless review of what happened.
12 min read - #AI#Kagent#Kubernetes
Diagnosing OpenCTI with Kagent (5) - Reviewing the Setup Against Google SRE
I reviewed the L402 payment gate's monitoring and on-call agent against the Google SRE Book and Workbook. Eight practices, what the review found, what changed, and the gaps that are still open.
9 min read - #AI#Kubernetes#Aperture
Diagnosing OpenCTI with Kagent (4) - Rehearsing Incidents with an LLM Agent
I rehearsed incidents against a Kagent on-call agent watching an Aperture L402 payment gate, then scored it against a playbook. The exercise found gaps in my own alerts too.
8 min read - #AI#Kagent#Kubernetes
Diagnosing OpenCTI with Kagent (3) - Designing Access Control in a GitOps Setup
Access control for kagent has two sides, who can change an agent and what a running agent can reach. How I split them across Git review, GitOps, RBAC, MCP, application, and database controls.
16 min read - #AI#Kagent#Kubernetes
Diagnosing OpenCTI with Kagent (2) - Tools, Code-First Diagnosis, and Read-Only Access
The four MCP tools behind the paid-service diagnosis agent, why the code classifies the state before the model explains it, and how Kubernetes access stays read-only.
5 min read - #AI#Kagent#Kubernetes
Diagnosing OpenCTI with Kagent (1) - Architecture: Agent, Model, and MCP Server
How I built a diagnosis agent for a paid OpenCTI and ASM service on kagent, and how the Helm values, Agent, ModelConfig, system prompt, and MCP server fit together.
5 min read - #Hoodi#Validator#DevSecOps
Prioritizing Security Controls: What Happened and What I Learned
Lessons from building a Hoodi validator on AWS about deployment speed, infrastructure cost, operational burden, and over-relying on AI when applying DevSecOps controls.
5 min read - #AI#AI Supply Chain#Supply Chain Security
The AI Software Supply Chain Is Under Attack
A field guide to AI-era supply chain attacks - the objects at risk, how to screen them, and why AI coding raises the stakes.
11 min read - #Security Requirements Plugin#Security Design#Security Requirements
Security Requirements Plugin: First Security Activity
Claude plugin that derives security requirements from service descriptions and reviews security design.
10 min read