Tag: security
All the articles with the tag "security".
-
Credential Sentinel: An Agent That Reads Widely and Acts Narrowly
Updated:How I built Credential Sentinel, a LangGraph AI agent that finds the credentials nobody is rotating and walks them to a safe, human-approved rotation. A look under the hood at the resumable state machine, the real TLS handshake, delayed-revoke cutovers, automatic rollback, and a cross-run memory.
-
Building the Agent Was the Easy Half. The Evaluation Wasn't.
How I evaluated my own AI agent: a 50-case golden dataset scored with code and an LLM-as-judge in LangSmith, plus the security regression I caught. The story of taking Credential Sentinel from a 0.976 composite score to a clean 1.0.