AI Security
Codex MCP setup: config.toml, enabled tools and approval modes
How OpenAI Codex stores MCP servers in config.toml, how to allow or deny tools, and which approval modes suit a security tool. Steps, table and checks.
Oct 19, 20266 min read
Resources
Supply Chain Security, in plain English.
Deep dives, practical guides, and incident analyses from engineers who build Safeguard. No fluff, no vendor FUD — just what you need to ship secure software.
Filter
Articles
RSS feedAI Security
Is a sandboxed coding agent safe? Myth versus fact
Most real risk with sandboxed coding agents is abuse of what you granted, not escape from the box. Six common sandbox myths checked against sources.
Oct 19, 20266 min read
AI Security
Where do secrets go in a cloud coding agent: setup or agent phase?
A cloud coding agent can treat setup scripts and the agent phase differently. See how one vendor documents secrets and caching, and what to test yourself.
Oct 19, 20266 min read
AI Security
Cloud coding agent or local agent: how does the risk change?
Moving a coding agent from your laptop to a cloud sandbox shifts risk rather than removing it. A side by side view of what gets better and what gets worse.
Oct 19, 20267 min read
AI Security
Claude file creation: which network egress setting should admins pick?
Claude's code execution and file creation has four network settings and plan based defaults. What the Help Center says and how a team should choose.
Oct 19, 20266 min read
AI Security
How do you put Claude Code MCP servers and hooks inside a sandbox?
Claude Code's built-in Bash sandbox leaves MCP servers and hooks outside. The sandbox runtime wraps the whole process. Setup steps, defaults, and limits.
Oct 19, 20266 min read
AI Security
What do Claude Code cloud session network access levels allow?
Claude Code cloud sessions offer None, Trusted, Full, and Custom network access. What each level allows, what bypasses it, and how to pick one for a repo.
Oct 19, 20266 min read
AI Security
Claude, ChatGPT, Claude Code and Codex: what is the difference?
A plain guide to the names you hear: a chat assistant answers questions, a coding agent edits files and runs commands. Why the security stakes differ.
Oct 19, 202610 min read
AI Security
Claude API MCP connector: what it supports and what it does not
The Claude API MCP connector reaches remote servers from the Messages API. Its limits, tool controls and token handling, for security tool use.
Oct 19, 20266 min read
Stay informed
Weekly insights on software supply chain security, delivered to your inbox.
Self-healing security runs on Safeguard.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.