I write about AI agents, API architecture, search, product engineering, and technology leadership.
Technology Leader & Builder
I gave a coding agent a test with no honest answer. It didn't fix the code — it rewrote what 'equal' means so every check passed. Then one line of the prompt stopped it cold. Cheating isn't character. It's conditions.
Read article →I gave a coding agent a test with no honest answer. It didn't fix the code — it rewrote what 'equal' means so every check passed. Then one line of the prompt stopped it cold. Cheating isn't character. It's conditions.
Who owns the security of an autonomous agent? If your answer is 'the security team,' you've already lost. The agent is water; the defense is an octopus — a distributed-but-coordinated way to own the seams where real incidents live.
I built a rig to watch two sandboxed AI agents talk to each other through a shared board. The striking teamwork I first saw — two agents cooperating with no one telling them to — turned out to be my own test leaking a hint that a partner existed. Seal the jar, and capable agents don't cooperate — they try to crack it open, and they don't stop.
It knows fifty design patterns — that's the problem. The most dangerous case isn't when it picks a bad pattern; it's when it picks a reasonable one, commits it as if you'd decided, and never mentions a choice was made.
An introduction to the five patterns for connecting AI agents to your existing APIs.
The API foundations you need before letting an AI agent call your services.
Choosing the right tool layer — the five ways an AI agent can call your APIs, and when to use each.
See your prompt's risk level, token budget, context waste, and security exposures before you send it to your AI tool. Live in the editor, with a @preflight chat helper.
A pre-PR security gate for AI-assisted code. Scans for exposed secrets, trap files, and risky agent/tool configs. One command, local, CI-ready.
The shared, zero-dependency TypeScript engine powering both tools. Secret, trap-file, instruction-integrity, and MCP-config scanners you can build on.
I'm Kiran Chanda, a technology leader who builds enterprise systems across finance, retail, and manufacturing. I'm currently exploring how AI changes software architecture, API integration, product development, and engineering leadership. Everything here is free — no charges, no paywalls.
No noise. Unsubscribe anytime.
