Blog/Tag
6 articles tagged Guardrails, newest first. Each one is sourced and written for people who ship.
Security & compliance
·13 min read
What to check after the model answers: safe Markdown rendering, schema checks before actions, moderation classifiers, streaming rails and groundedness.
Security & compliance
·10 min read
A review of Guardrails AI: 65 hub validators, eight on-fail actions, the August 2026 shutdown of hosted inferencing, and when NeMo Guardrails fits better.
LLMOps & evals
·11 min read
Portkey: an OpenAI-compatible gateway for retries, fallbacks, caching, guardrails and cost tracking. Its config, latency cost and when to self-host.
AI agents
·10 min read
A review of the OpenAI Agents SDK: the runner loop, tracing, guardrails and approvals, plus what the release churn and the Responses-only features cost.
Security & compliance
·9 min read
A review of Lakera Guard: one endpoint before the model, the PINT benchmark behind its scores, and why the free tier stops at 10,000 requests a month.
Security & compliance
·9 min read
Rebuff scores prompts with heuristics, an LLM, a vector store of past attacks and canary tokens. Archived in May 2025; last release January 2024.