RepoRadar
Leave a Tip
Item detail

secureagentics/Adrian

Adrian is an Apache-2.0 runtime security monitoring and control engine for AI agents, AARM-aligned (the open Agentic AI Risk Management standard). 310 stars, Python SDK with a two-line install to LangChain agents, analyzes agent activity logs (tool calls, actions, outputs) and reasoning traces to detect malicious, misaligned, or out-of-remit behavior, and can intervene in-flight.

Open source →Back to radar
Score7.6
Popularity68.0
Riskconditional
TierSilver
Score breakdown
Usefulness7.0
Novelty8.0
Momentum6.0
Maturity7.0
Open-source/build8.4
Evidence7.2
Workflow potential8.4
Setup ease8.8

Popularity is tracked separately. Support, ads, sponsorships, and tips never affect these signals.

Why it matters

Useful for platform and security teams who already run LangChain agents in production and need a runtime observability + control layer to catch prompt-injection, rogue tool use, or out-of-scope behavior without rewriting the agent.

Who should use it

platform teams running LangChain agents in production who need runtime safety netssecurity teams that need to detect prompt injection, rogue tool use, or out-of-scope agent behaviorcompliance teams that need AARM-aligned audit logs of every agent tool callagent framework authors adding runtime policy enforcement as a drop-in layerteams that need in-flight intervention (block / modify / alert) rather than just observability

Who should skip it

Skip if the source link, docs, or setup requirements do not match your workflow.

Risk explanation

Sits in the tool-call path of every agent you wrap — treat Adrian as a credentialed service; 310 stars as of 2026-06 — early-stage project, validate against your threat model before depending on it for production safety; Behavioral policy is only as good as the rules you write — Adrian detects what you tell it to detect.

Evidence links

Closest alternatives / related signals

agent-securityruntime-monitoringaarmlangchainprompt-injectionpolicypythonapache-2