Thoughts on software architecture, AI/ML systems, production engineering, and building products as a solo founder.
A null-prompt ablation across 19 agent configurations: 13 survive the noise floor, two of them scoring zero without their prompt; four say more about the tests than the prompts. Anthropic runs the same kind of ablation on Claude Code and reports the opposite direction. The difference is scope.
An HTTP 200 means the request came back, not that the work happened. A rotated token, an alerting path that went silent for days, and why the same trap gets worse once agents are wired to real tools.
Multi-LLM routing looks clean on a whiteboard. In production the real risk is the cost math that hides its own downside, latency treated as a single number instead of a distribution, and the silent failures that return a clean HTTP 200.
Why AI agents wired to real tools through MCP should default to deny, and how to build the per-operation, audited RBAC that makes fail-close real.
What survived contact with real usage maintaining an MCP server across three clients I never built for: fail-close defaults, replayable observability, and tool descriptions as the real interface.
The pre-mortem is the cheapest way I know to catch the preventable failures before they ship. The 25-minute version I actually run, and when not to bother.
A free MIT CLI that scans a repo for seven classes of AI-agent-gateway anti-patterns — tool-access/RBAC, fail-close, supply-chain pinning, observability, multi-LLM routing, secrets, production-readiness — runs as a CI gate, and exposes itself as an MCP server.
When one team adds a request header, the backend CORS allowlist becomes a coordination point. Here is what that tax actually costs, why it fails silently, and four patterns for handling it.
Four runtime events, four policy-attachment surfaces, and the hook coverage at each event is what separates a demo from something a regulated buyer will sign for. Framework selection is a distraction.