Practical notes from building and operating LLMOps and AI agent systems.
I sent the confirmed priority request from part two into an isolated environment created from a fixed starting point. The saved AI proposal from part one failed three of nine repository tests and the priority-filter acceptance check, but one predefined repair passed four check groups and produced a local commit without changing the remote GitHub repository.
I built an interface that refuses to send ambiguous requests such as 'add priorities and make Todos easier to read' directly into implementation. It records answers, manual requirement edits, and human confirmation, then verifies that the server blocks change jobs while questions remain unanswered or the requirements remain unconfirmed.
I built a dedicated web interface for requesting Todo app changes, normalized responses from Sakura AI Engine, Cloudflare Workers AI, and Google AI Studio, and tested whether a seven-file code proposal could stop after failing three of nine fixed tests without changing GitHub or production D1.
I sent OpenTelemetry spans from Cloudflare AI Gateway and a Node.js agent to Grafana Tempo under the same trace ID, then compared a successful response, a client timeout, an invalid model error, and a fallback. The results reveal missing error spans, different duration boundaries across the app, Gateway logs, and OTel spans, and the need for a separate observation point for response quality.
I use API cost, latency, evaluability, and deterministic alternatives as pre-implementation gates, then apply them to fixed scenarios to decide which tasks should not enter a generation-evaluation loop.
Inspired by Kiro's requirements analysis, I ran a minimal verification that turns contradictory acceptance criteria and undefined exception paths into clarification questions before code generation.
Using Cognition's SWE-1.7 article as a starting point, this article verifies what an AI-driven development resume note needs in order to restart long-running work, using a local verification script that does not call an external API.
A verification log on normalizing Langfuse Observations and Scores into ClickHouse, then using SQL to inspect model-level cost, prompt-version quality trends, and retry-driven cost increases.