Technology

Technology

A Pending Approval Made My Release Gate Skip Its Install Audit

August 7, 2026 • 8 minutes read • 69

A pending approval put an entry in the error list, and one guard clause then skipped my installer's real install audit. The gap serialized as an empty object. Here is the fix.

Read story

Technology

We Wired ChatGPT and Gemini as Tool-Connected Browser Workers — and One of Them Said No

July 9, 2026 • 5 minutes read • 68

A builder-in-public field note on wiring ChatGPT and Gemini as tool-connected workers in a real browser tab — why Gemini complied, why the ChatGPT account refused, and why framing decides everything.

Read story

Technology

Preventing Permission Drift: Why Agent Evaluation Passing Does Not Equal Action Authorization

July 4, 2026 • 6 minutes read • 97

An AI agent can pass every eval and still take an action it was never allowed to take. Permission drift is a control-plane failure, not a model failure. Here is the fix.

Read story

Technology

The Day Skynet Stopped Calling a Tool Result Proof

July 3, 2026 • 5 minutes read • 131

A builder-in-public story about replacing tool-reported success with independent live-page, archive, image, and signature proof in Skynet.

Read story

Technology

Directed AI Video Needs an Audio Ledger

June 6, 2026 • 6 minutes read • 95

A Skynet Blog field note on why directed AI video belongs in the operating record, how Platform differs, and how audio-video proof gates work.

Read story

Technology

Why Upgrading Foundation Models Will Not Fix Multi-Agent System Coordination Failures

June 4, 2026 • 8 minutes read • 116

Why a smarter LLM will not fix an AI system that fails at orchestration and verification: reliability comes from the weakest coordination layer, not the model.

Read story

Technology

Central Orchestrator Failures in Multi-Agent Systems: Separating Coordination from Worker Liveness

June 3, 2026 • 12 minutes read • 124

A 2026 field note on why multi-agent AI fleets fail at the control plane, and how live probes, decoupled agents, fallback chains, and independent monitors keep capability alive.

Read story

Technology

What Does One AI-Assisted Platform Post and Blog Post Cost With Codex?

June 1, 2026 • 5 minutes read • 71

A transparent Codex credit estimate for producing one Platform post and one Blog post, with official OpenAI rate-card inputs and clear limits.

Read story

Technology

Managing Intent Across Multi-Tool Control Loops: High-Throughput Execution After Eliminating Drift

May 25, 2026 • 6 minutes read • 113

A source-backed reflection on Skynet, anti-drift AI orchestration, provider routing, TODO memory, CDP proof, and execution speed.

Read story

Technology

Engineering Validation of the Philippine Food Price Dashboard Upgrade

March 18, 2026 • 6 minutes read • 150

Published: March 18, 2026 • Part of the Philippine Food Price Prediction series This post documents the upgraded system as deployed on March 18, 2026. Engineering validation companion This Blog page documents the March 18, 2026 system upgrade, test coverage, implementation architecture, alert pipeline,

Read story

Technology

Forecasting Philippine Food Prices: Evaluating Machine Learning Models Against Real Out-of-Time Data

March 5, 2026 • 20 minutes read • 323

I wanted to know whether two decades of Philippine food-price data could produce a forecast that survives a real holdout period, not just a clean training score. I trained the Random Forest on 2000-2019 data and held out 2020-2023. It reached R² 0.91, MAE

Read story