AGENT SECURITY
12 items
12 posts
GitGuardian counts 31,088 secrets in Claude Code settings files on public GitHub, 5,343 valid, plus copies baked into Docker Hub images. What its numbers add to Apiiro's, and the Monday checklist.
Goldman Sachs limited Claude Code and GitHub to engineers in October. What Bloomberg reported, what it signals for banks and regulated teams, and a rollout checklist for your own org.
Apiiro found real secrets in 1,122 public Claude Code settings.local.json files saved by Yes, and don't ask again. What the file stores, what Anthropic said, and a 10-minute cleanup.
SneakPeek (CVSS 8.7) let a poisoned HTML file read SSH keys and cloud credentials via the Claude Code Desktop preview pane. Fixed in Claude Desktop 1.30096.1.
A guard hook that crashes, times out or points at a missing script lets the tool call through by default in Claude Code. v2.1.295 adds onFailure: block. Here is what each coding agent does when your policy hook breaks, with real test runs.
Copilot CLI 1.0.94 finds local Ollama models from /model and ships an OS-level sandbox. Setup, the 4k context trap, and why local is not offline.
A deny rule on .env stops cat but not a Python one-liner. Here are the three layers that actually keep secrets out of a coding agent's context - file rules, environment scrubbing and credential masking - and what Claude Code, Codex, Cursor and Gemini CLI give you for each, with real test runs.
Google's Gemini agent can create coworker agents with their own Workspace account, email and audit trail. What Google announced and what builders should copy.
Claude Code mods are plugins whose JavaScript handlers run inside Claude Code. What they do that hooks and skills cannot, a first mod, and the trust question.
OpenAI Dots turns ChatGPT into a long-running work agent. For developers, the real question is where approvals, permissions and sandboxes sit.
Cloudflare AI Gateway now ties requests to a verified identity and learns per-account baselines. How the anomaly math works and why it beats global thresholds.
Microsoft Execution Containers (MXC) sandbox AI agents with JSON policies on Windows, Linux and macOS. The TypeScript SDK, backends and setup.

Get Smarter About AI Dev
New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.