1. Conduid
  2. Agents
  3. Safetydrift
MCP server · Agents

Safetydrift

by · github.com/msabhishek0820-prog/safetydrift

First open-source implementation of the SafetyDrift framework (arXiv:2603.27148) — predicts when individually safe AI agent actions compound into a safety violation using Markov chain analysis. Works as an MCP middleware.

Unclaimed agentic-aiai-agentsai-safetydrift-detectionguardrailsllm-agentsmarkov-chainmcpmodel-context-protocolpython
InstallView source
34Low

Scored 28 days ago · breakdown

About Safetydrift

Safetydrift is an MCP server published by msabhishek0820-prog in the Agents category: first open-source implementation of the SafetyDrift framework (arXiv:2603.27148) — predicts when individually safe AI agent actions compound into a safety violation using Markov chain analysis. Works as an MCP middleware.

The repository has 1 stars and 0 forks.

Install

From source
git clone https://github.com/msabhishek0820-prog/safetydrift

This server has no ConduID identity, so agent calls to it are not receipted. Pin the version you install and review the source before granting it credentials.

Ask AI

✦

Ask AI about Safetydrift

Powered by Claude · Grounded in docs

I know everything about Safetydrift. Ask me about installation, configuration, usage, or troubleshooting.

Security checks

6 checks have not been run yet.

Questions

About Safetydrift

How do I install Safetydrift?

No package has been published for this server, so install it from source: clone the repository and follow its README.

Is Safetydrift safe to use with an AI agent?

Its trust score is 34 out of 100 (low). Conduid hasn't run static security checks on this repository yet, so review the source yourself before granting it credentials. It has no ConduID identity yet, so agent calls to it are not receipted.

Is Safetydrift still maintained?

Conduid hasn't recorded a commit date for this repository yet. Check the repository directly for recent activity.