Free tools
AI hardening kit
Policy files for the AI coding tools your developers run, generated in your browser. Every key comes from the vendor's official documentation and links to it; options we could not verify are left out. Nothing you type or pick is sent to ThreatDeck.
- Claude Code managed settings managed-settings.json · settings.json Block secrets and credential files, disable bypass mode, sandbox Bash, lock permission rules, MCP servers and hooks to managed policy, set transcript retention and telemetry. Strict, Balanced and Developer-friendly presets. Open the generator
- MCP allowlist Claude Code · GitHub Copilot / VS Code · Cursor · JSON / CSV Pick approved MCP servers from MCP Server Watch or add internal ones, see advisories and risk signals for each, and copy version-pinned allowlists for every client. Open the generator
- GitHub Copilot managed settings Enterprise managed-settings.json Permission rules for secrets and risky commands, bypass and assisted modes, computer use, a sandbox floor for CLI, VS Code and the Copilot app, and the MCP allowlist. Open the generator
Not generated (yet)
- Cursor: security controls (Privacy Mode, MCP allowlist, model and repository restrictions) are set in the team dashboard rather than a policy file; the MCP allowlist produces entries to paste there. Cursor hardening guide
- GitHub Copilot content exclusion is configured in repository and organisation settings on GitHub. Content exclusion
Before you deploy: check the minimum safe versions of each tool, and read the playbooks Secure AI coding tools and Lock down MCP.