@raplsworks

Before you disconnect MCP servers, read the context breakdown

12h ago · 4 min read · When an AI coding agent starts drifting mid-session, forgetting a constraint, repeating a solved question, giving thinner answers than it did an hour ago, the temptation is to reach for the explanatio

Prompt injection isn't the whole story: secure what the model hands back

1d ago · 5 min read · Here is the rule I wish I'd started with: the output of an LLM is untrusted input. Not "mostly clean because a model wrote it." Untrusted, the same as a string a user typed or a response that came bac

A WordPress chatbot that answers from your own content: the design decisions behind it

2d ago · 6 min read · I built a free WordPress plugin called Rapls AI Chatbot, so treat this as design notes from the person who made the calls, not a neutral review. The useful part here is the reasoning behind a few deci

Conversation or straight line: how I split work between Claude Code and Codex

3d ago · 6 min read · I keep two agentic coding CLIs running: Claude Code and Codex. For a while that felt like I couldn't pick one. What fixed it wasn't picking one. It was finding a single question that tells me which to

Claude Fable 5 lasted three days. Then the US government pulled it.

4d ago · 13 min read · On Tuesday this week I was reading launch coverage that told me to try Claude Fable 5 soon. By Friday night it was gone. Not deprecated, not rate-limited, not behind a waitlist. Gone, by order of the