#agentsecurity

One coding agent answered repo-controlled git config with a confirmation prompt in August. Another...

The MCP tasks extension requires an authorization check on every task request. Its own rationale says...

HOL Guard 3.0 puts PreToolUse on a version-matched Rust path with no Python semantic fallback, makes Local Extensions the capability vocabulary for managed-restrictive Control Sets, and keeps sync pau

Spring AI released a fix on 21 August for a bug where a tool absent from the request could still be...

The Problem Is Not the Model. It's the Missing Gate. In 2026, autonomous pipelines are no...

You left Claude Code running. It tried to send your AWS keys to a website you never opened. Guard froze it: Allow just this once, or Keep blocked.

You lock down every other part of your stack with zero-trust. Your AI shouldn't be the exception. Now...

Coverage of the reasoning-trace harvest reported three different totals as if they disagreed. They...

The gym booking story is being told as an API with no authorization. The agent's own account says two...

HOL Guard 2.1 ships a 51,000-case command classification corpus, HMAC-backed dashboard reconnect, signed Codex hook manifests, DNS-pinned archive downloads, and a typed GitHub capability model — all o

We Built a CCS-Native MCP Server Today we launched correctover-mcp-server — an MCP server...

Prime Intellect's new agent beat the human baseline on ARC-AGI-3 at launch. I copied one idea, queued one benchmark, and installed nothing.