Guides
Agents, MCP and what changes when software signs in on someone's behalf.
The engineering questions that matter are rarely about which tool.
The uncomfortable part of building for agents is that the old assumptions quietly stop holding.
The useful frame is not "is this secure" but "what does an attacker get if this one thing goes wrong, and how quickly would we know".
The value of a standard is that two parties who have never spoken can interoperate.
Most of the difficulty with agents is not the model.
Almost every incident worth writing about starts with something that was working exactly as designed.
Protocol work looks tedious from the outside and turns out to be where the interesting failures live.
A design is a set of bets.
Agents changed the shape of the problem.
The gap between selling to teams and selling to companies is not a pricing change.
Most systems do not fail because someone chose the wrong database.
Every pattern here is already implemented in the platform.