Blog
Real-world notes on software engineering — architecture, frontend, backend, AI, DevOps, and everything in between.
Fabio Borges
September 7, 2026
OpenCode Go now requires x-opencode-session — here's why and how to fix it
OpenCode Go started enforcing the x-opencode-session header for routing optimization. If your chat widget or agent broke today, here's what happened and how to fix it.
Fabio Borges
September 3, 2026
Mysterious Contributor on my GitHub Repository: What Happened and How to Avoid It
A contributors pill shows three avatars, the API lists one, and git log agrees with the API. We traced a live incident on atlaslink — branch fix/issue-7-isolate-session-diagram showed fworks-tech, claude and krkshw — to a force-pushed orphan history, GitHub's two contributor graphs, and the cache that makes ghosts linger for weeks. Here is which graph to trust and how to prove it.
Fabio Borges
September 3, 2026
Never Leave the Chatbox Hanging: Fixing a Silent Timeout in Our AI Assistant
A real visitor question about the Atlaslink project froze our portfolio AI — Thinking, then Typing, then nothing. No error, no recovery. Here's the four-layer postmortem and why every layer of an AI feature needs its own way to fail fast.
Earlier posts
Fabio Borges
August 11, 2026
The Agenthood news digest now writes itself
A scheduled AI agent drafts the Agenthood news digest from GitHub releases, validates it against a strict contract, and opens a PR for human review. Here's how it works and what our own review caught.
Fabio Borges
August 9, 2026
DevSprint — JavaScript Quiz Mini-Game
Test your JavaScript fundamentals with DevSprint: 20 questions, 3 lives, 20 seconds each. Leaderboard, achievements, and referral opportunities for top scorers.
Fabio Borges
August 4, 2026
Effective Prompt Techniques: 6 Foundational Prompting Techniques
We explain that prompt engineering is simply the practice of designing effective instructions for AI systems, combining familiar human communication principles with AI-specific considerations. We introduce six foundational techniques: giving context, showing examples of desired outputs, specifying constraints, breaking complex tasks into steps, asking the AI to think first, and defining the AI's role or tone. We also share troubleshooting strategies for when responses aren't quite right and highlight common patterns that lead to successful interactions.
Fabio Borges
August 2, 2026
How to Build an AI Agent: The Complete Engineering Guide
A practical, step-by-step guide to building production AI agents — from purpose definition to testing. No hype, just the engineering decisions that matter.
Fabio Borges
August 1, 2026
The Bouncer: how we defend a public AI endpoint from prompt injection, cost drains, and IP spoofing
flabs.tech's AI assistant is a public, unauthenticated endpoint that calls a paid model. We built a deterministic abuse-prevention pipeline — investigation, quarantine tiers, two-tier injection detection, and shadow-mode rollout — without ever locking out a legitimate visitor.
Fabio Borges
July 31, 2026
DeepSeek V4 Flash on OpenCode Go now requires China hosting opt-in
The official V4-Flash-0731 release (July 31, 2026) is served exclusively via DeepSeek's PRC infrastructure. OpenCode Go/Zen now requires an explicit 'Enable models hosted in China' toggle. Here's what changed and your options.
Fabio Borges
July 21, 2026
Your Portfolio Should Answer Questions, Not Just List Them
We added a floating AI assistant to flabs.tech that knows everything about Fabio's experience, skills, and projects. Here's why every portfolio should have one.
Fabio Borges
July 6, 2026
ChainTelescope — from Jupyter notebook to production crypto command center
How a personal crypto analysis notebook evolved through two renames and a full modular rewrite into a production Streamlit dashboard with multi-provider fallback, AI assistant, and CI-enforced quality gates.
Fabio Borges
July 6, 2026
OpenCode Zen and Go — what they are, what they cost, and how Agenthood connects to them
OpenCode is more than an open-source coding agent — it also runs Zen (pay-as-you-go) and Go (subscription) model APIs. Here's how they work, what they cost, and how Agenthood integrates with both.
Fabio Borges
June 26, 2026
Agenthood v3.0.0 — M5 Intelligence is shipped
v3.0.0 introduces hierarchical RAG, agent-controlled retrieval, 16-member orchestration, OpenCode providers, governance docs, and 527 passing tests.
Fabio Borges
June 17, 2026
Agenthood is now on SkillsMP and Skills.sh
The full Society now ships as open Agent Skills — one canonical SKILL.md each, installable from Skills.sh or browsable on SkillsMP.
Fabio Borges
June 13, 2026
Agenthood: The Academy is live
25 practical articles across 3 levels — from GenAI and RAG fundamentals to production-ready agent systems.
Fabio Borges
June 5, 2026
GraphQL Federation at scale — what I learned shipping it to a US telecom
Shipping Apollo Federation to production for a major US telecom's AI checkout flow — subgraph architecture, N+1 elimination, WunderGraph as the gateway, and the observability setup that let us actually debug it.
