Open-source guardrail framework from Meta that runs layered scanners around agent execution: PromptGuard 2 for jailbreak detection, Agent Alignment Checks that audit chain-of-thought for goal misalignment, and CodeShield static analysis of generated code. Meta describes the alignment auditor as still experimental.
open_source · generally available · Research snapshot 2026-09-06
Show me LlamaFirewall scanning an email agent's inputs with PromptGuard 2 and flagging an indirect injection through Agent Alignment Checks, including the experimental caveats.
Capabilities and evidence
Support labels reflect the supplied research. Documentation and vendor claims are not independent product tests. “Not established” means the researcher did not find support; it does not prove a capability is absent.
Documented by provider
LlamaFirewall provides PromptGuard 2 as a jailbreak detector, Agent Alignment Checks as a chain-of-thought auditor for prompt injection and goal misalignment, and CodeShield as an online static analysis engine for insecure generated code.
Limit: Published as research and open source; production support commitments are not stated.
The framework is designed for integration into existing AI agents and LLM applications, with scanners configured per role and invoked on input at runtime.
Limit: The documentation page fetched shows prompt-guard scanning only and does not demonstrate tool-call or output enforcement.