AI coding guides & resources
Engineering decisions, tutorials, and video guides from the OTF team.
Codex vs Claude Code: why the model is not your scaling bottleneck
Codex vs Claude Code: features don't scale you — context windows bind both, so codebase clarity beats model choice every time.
6 min read
Introducing Package Intelligence to Manage React Native and Expo Dependencies
Upload your package.json, get health, Expo compatibility, and upgrade guidance in one dashboard. What the launch confirms.
5 min read
Cursor's new Jira integration simplifies developer workflows with AI ticket handling
Cursor's Jira integration routes tickets straight to its coding agent. Setup costs, the four-ticket clarity test, honest limits, and how to start today.
6 min read
MiniMax M3: million-token context just got cheap — here's how to put it in your agent
MiniMax's M3 teases sparse attention with up to 15.6x faster long-context decoding. Here's how to wire the M-series into your agent now.
6 min read
The real cost of vibe coding: why AI agents drain budgets and stall launches
Vibe coding with AI agents feels free until token bills, sandbox rewrites, and retries stall your launch. Here is the real cost breakdown.
8 min read
Lovable AI backend architecture: what it does and what you still own
A source-checked guide to Lovable’s backend options, Supabase connection, migrations, secrets, deployment, and export boundaries.
9 min read
Visa-backed agent payments in Replit: why AI builders should care about real spend
Replit's Visa-backed identity layer lets agents spend real money. Cost-abstraction failure goes from annoyance to invoice liability.
7 min read
Dependency bloat in React Native: why AI tools can't solve your upgrade pain
AI scaffolds React Native apps in minutes, but dependency rot decides if you survive the next Expo upgrade. Own it.
8 min read
AI agents broke jqwik: when vibe coding meets a maintainer's kill switch
A maintainer reportedly hid a prompt-injection trap in jqwik 1.10.0 — what AI builders must learn.
8 min read
Why AI-coded projects feel disposable, and how to build code you trust
Why AI-coded projects feel disposable — sandbox debt, missing context, rented code — and how ownership fixes it.
7 min read
AI code you cannot reuse is not worth the token bill
Throwaway AI code costs tokens and trust — own reusable code instead.
8 min read
Claude Code's dynamic workflows: parallel agents hit the context wall fast
Dynamic workflows parallelize Claude subagents but can't escape sandbox and context limits.
7 min read
Claude Code's tool pipeline: why agent security defaults are a production wall
Claude Code's tool pipeline keeps agents safe by default — and that same fail-closed design blocks hands-off production automation.
7 min read
OTA updates in React Native: the fix for when App Store review ruins Friday
App Store review is an unplannable SLA. OTA updates patch React Native JS without review — within a hard boundary.
6 min read
Most Claude Code output is throwaway until you own the workflow
Agent-written code stays disposable until reviewed, tested, and integrated — how to make output you own.
5 min read
Rork can scaffold a React Native app. It can't follow you to the App Store.
Rork scaffolds React Native apps fast and now offers a publish path. But store submission still runs on your accounts, keys, and review.
7 min read
Your AI IDE changed its billing, its price, and its owner — and you didn't get a vote
Windsurf changed its billing model, raised Pro 33%, and got acquired in one quarter. Own your code so rented tools stay replaceable.
7 min read
Your design system is the context your AI agent is missing
Encode design decisions as semantic tokens plus a lint rule that bans raw values, so AI agents can't drift.
7 min read
'Just export it' is the most expensive advice in vibe-coding
Exporting vibe-coded apps hands your agent files without history or conventions, so it invents instead of extending. Here is the real cost.
7 min read
Uniwind and Gluestack put shadcn on React Native, but you still copy-paste per platform
Uniwind and Gluestack bring shadcn to native, but web and native stay two registries with no shared API contract.
6 min read
Expo SDK 56 killed the platform-specific file — one component renders on iOS, Android, and web
Expo SDK 56 makes universal components production-ready — one file on iOS, Android, and web, plus what to verify before adopting.
6 min read
Your vibe-coded app has no View, so it cannot go mobile yet
Your Lovable app has divs, not Views — the structural reason web output can't become a native app.
7 min read
Notion just made Claude Code a tracked teammate — Lovable wasn't invited
Structural analysis: Notion's connection model favors file-system agents over sandboxed builders — and what that means for your stack.
6 min read
Replit Agent 3 turned a $25 plan into a $1,000 week — the credit pool nobody reads
Usage-based AI builder pricing hides many meters behind one credit pool, and the agent loops hardest exactly when it bills most.
7 min read
