Writeups
Patterns from building agent systems, each with code or configuration and without client references.
- Running agents in production: switch off, measure, swap Whether an agent is ready for production depends on the checking and operating architecture around it, not on the choice of model. Seven patterns for it, from the trust model and a stateless confirmation to kill switch stages, evals as a release gate and events without values.
- When the agent can write: securing write access An agent that changes data needs declared write operations, an approval that shows the change in plain text, and a tamper-evident log. The server enforces all of this; the prompt does not.
- How long Python stacks keep trusting a compromised certificate authority Browsers block a compromised certificate authority within hours. Python services keep trusting it until a package or the interpreter has been updated in every venv, image and lockfile. Using the eMudhra case, I show where the bundle lives and how to check it.
- Exit readiness is a security property A locked account, a failed region or access by a foreign jurisdiction hits a system as hard as a bug in the code. How I inventory and assess a system's dependencies on its vendors, and what is different with agents.
- The LLM gateway is a secrets manager To use someone else's model access, attackers do not need to attack a model. A key is enough, and keys are concentrated in gateways, platforms and logs. The paths attackers take to get there, and how I secure a gateway against them.
- The BSI baseline defense against indirect prompt injection as a running RAG chat Germany's Federal Office for Information Security (BSI) has published five baseline measures for document-based LLM chats. I built them into a small RAG chat in Python, ran it against a real model and show where agents need a permission and approval layer to take over.
- MCP attack patterns, mapped to OWASP Agentic 2026 Tool poisoning, rug pulls, token passthrough and overly broad tools: how the known MCP attacks work, where OWASP places them and which countermeasures can be expressed in code and CI.