Aviral Srivastava·Oct 23I Found Two OAuth Security Flaws in the Official MCP SDKs.How my research into the official MCP Python and TypeScript SDKs uncovered OAuth credential exposure vulnerabilities, now tracked as…
Aviral Srivastava·Aug 12Three CVEs in Activepieces: The Sandbox Was Real. The Code Just Didn't Run Inside It.A command injection that fires before any sandbox exists, a V8 isolate that protects the wrong half of your module, and an XSS in the OAuth…
Aviral Srivastava·Aug 5CVE-2026–69258: Flowise Patched overrideConfig. I Found the Two Places the Patch Never Reached.An unauthenticated property injection into the flow execution context of any public chatflow. The gate was already in the codebase. It just…
Aviral Srivastava·Jun 29CVE-2026–53923: How a 32-Bit Integer in vLLM Leaks One User’s GPU Memory Into Another’sIn multi-tenant LLM serving, the bytes your model reads might not be yours. One wrong type in a CUDA kernel is all it takes.
Aviral Srivastava·Mar 29I Found 5 Security Vulnerabilities in XGBoost. Here’s What Happened.XGBoost is one of the most important libraries in machine learning. 26,000+ GitHub stars. Used by banks for fraud detection, insurance…A response icon1A response icon1
Aviral Srivastava·Mar 19CVE-2026–33017: How I Found an Unauthenticated RCE in Langflow by Reading the Code They Already…In early 2025, CISA added CVE-2025–3248 to their Known Exploited Vulnerabilities catalog. It was an unauthenticated remote code execution…A response icon1A response icon1
Aviral Srivastava·Mar 15I Found a SQL Injection in an AI Agent.How a 56,000-star open source LLM app forgot to parameterize its queries, and why that matters way more than you think.
InSystem WeaknessbyAviral Srivastava·Feb 26AdversaryPilot + garak: How to Turn LLM Scan Results into a Next-Step Attack StrategyYou ran garak. You got a JSONL report. Now what?
Aviral Srivastava·Sep 8, 2025Schema Smuggling: Breaking AI Tool-Calls with Parser QuirksIf your agent “trusts JSON from the model,” I don’t need admin — I need your parser to blink for one millisecond.A response icon1A response icon1
Aviral Srivastava·Sep 3, 2025Batch-Neighbor Exfiltration (MoE × SpecDec): Turning Shared Inference Into a Keyword OracleYou don’t need model access to leak secrets — just be your victim’s batch neighbor.