Run a free red teaming scan on your AI agent
Saish Bhorpe
Author

Saish Bhorpe

AI Security Researcher, Repello AI

Posts by Saish Bhorpe

6 posts

Name-keyed MCP approvals in Claude Code: a project's MCP approval binds to the server name, not the command you reviewed.

A Trusted Name Is Not a Trusted Command: MCP Approvals in Claude Code

Jun 10, 20269 min read
Workspace Trust Is Not AI Consent: how a VS Code workspace setting can start Copilot Chat in Bypass Approvals mode.

Workspace Trust Is Not AI Consent: A VS Code Copilot Bypass-Mode Case Study

Jun 4, 20269 min read
AI threat modeling adapts STRIDE and PASTA for LLM-specific risks. How to map the 5 AI attack surfaces, document threats, and validate with red teaming.

AI Threat Modeling: How to Map the Attack Surface of Your LLM Applications

Apr 1, 20268 min read
The LiteLLM supply chain attack: how TeamPCP backdoored the AI development ecosystem

The LiteLLM supply chain attack: how TeamPCP backdoored the AI development ecosystem

Mar 26, 20267 min read
AI Adversarial Attacks: Types, Examples, and Defences

AI Adversarial Attacks: Types, Examples, and Defences

Mar 26, 202612 min read
ClawBot's GitHub Actions Attack: Supply Chain Worm, Kubernetes Wiper, and What AI Pipelines Need to Do Now

ClawBot's GitHub Actions Attack: Supply Chain Worm, Kubernetes Wiper, and What AI Pipelines Need to Do Now

Mar 23, 202614 min read