Raed's blog

https://blog.raed.dev

8 posts · 18 Votes

Tech 88% · Life 13%

Subscribe via RSS

  1. Why is your AI using so many tokens?

    A scroll-driven walkthrough of LLM context and how tokens pile up.

    0 ▲
  2. It wasn't the AI that flipped Proton's charge

    Proton's CEO called a request for security testing "loser thinking". Using AI to write code is fine, but for a privacy company, the guardrails around that code are the product.

    17 ▲
  3. What an adult ADHD diagnosis taught me

    For most of my life I called it laziness. A late ADHD diagnosis didn't fix my brain, but it changed how I talk to myself on bad days.

    1 ▲
  4. Microsoft Teams is a red flag

    Teams looks free because it comes in the bundle. The people paying for that choice are the ones stuck using it every day.

    0 ▲
  5. Code review assumes an author

    Code review assumes the person who opened the PR understands the code. Agentic development breaks that assumption. A look at why, not a fix.

    0 ▲
  6. Claude Code doesn't trust Claude with permissions

    How Claude Code enforces permissions with deterministic code, not model inference. The one LLM-based path is internal-only, fail-closed, and stripped from the public package.

    0 ▲
  7. Claude Code is a vibe-coded mess. Some of it is actually good.

    Some ideas worth stealing from the leaked Claude Code source: deferred tool loading, diminishing returns detection, cache-aware context compaction, and coalesced background extraction.

    0 ▲
  8. LangGraph's HITL Has a Double Execution Problem

    LangGraph's interrupt/resume primitives are genuinely clean. But the checkpoint model has a side effect that silently double-executes tools, and approval binding is positional rather than explicit. Here's what that means in practice.

    0 ▲