# Read-only, restricted, contained: the agent guardrails that failed open.

Source: [DEV Community: ai](https://dev.to/humanbound_ai/read-only-restricted-contained-the-agent-guardrails-that-failed-open-bj1) · 2026-11-21 · AI
Page: https://themorning.dev/story/read-only-restricted-contained-the-agent-guardrails-that-failed-open-4b4063750fd0b860
Kind: security · Impact: medium

## In brief

- On 19 and 20 November, crewAI, OpenClaw and LiteLLM each published a security advisory.
- Each advisory involves a guardrail that existed and was switched on, such as a read-only SQL mode or a deny policy.
- Humanbound's engine and CLI are open source, at github.com/humanbound/humanbound.

## Why it matters

Enabling a built-in guardrail did not reliably contain agent behaviour in these frameworks, so developers should not rely on a single switch to restrict what an agent can do.

## Timeline

- 2026-11-19: crewAI, OpenClaw and LiteLLM published security advisories (19 and 20 November)

## Coverage

- [Read-only, restricted, contained: the agent guardrails that failed open.](https://dev.to/humanbound_ai/read-only-restricted-contained-the-agent-guardrails-that-failed-open-bj1) — DEV Community: ai, 2026-11-21

Tags: ai-agents, guardrails, security, llm, vulnerability

---

Brief generated by AI (claude-haiku-5-5) from the reporting listed under Coverage. Headlines belong to their publishers. The Morning Dev – https://themorning.dev
