Claude Code permission modes compared: default to bypass
Six Claude Code permission modes, what each runs without asking, how to switch with Shift+Tab, and which to use for review, planning, CI and containers.
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.
Six Claude Code permission modes, what each runs without asking, how to switch with Shift+Tab, and which to use for review, planning, CI and containers.
Claude Code telemetry is off until you enable it, and prompt text is redacted by default. Which switches add prompts, tool details and raw API bodies, and who controls them.
Local, project and user scopes decide who sees an MCP server and where its config lives. Learn which to use, and how precedence and approval work.
How to allow, ask or deny MCP tools in Claude Code: the server and tool naming, wildcard limits, and why deny rules always win over allow rules.
Managed settings sit above user and project settings. File paths per OS, server-managed and MDM options, and a few keys that lock login, providers and modes.
How to run Claude Code non-interactively in scripts and CI: the print flag, bare mode, allowed tools, dontAsk mode and JSON output.
Hooks run your script at fixed points in Claude Code's lifecycle. How PreToolUse hooks block with exit code 2, how decisions combine, and what exit 0 means.
Set up the Claude Code GitHub Action with least privilege: secrets, app permissions, who can trigger runs, and cost caps, based on Anthropic's docs.
Claude Code checkpoints let you rewind file edits and conversation, but not Bash changes, most subagent edits or outside edits. The limits and the safe habit.
Weekly insights on software supply chain security, delivered to your inbox.
Your first fix PR is minutes away.
No sales call required, even your agent can complete the purchase over MCP.