How to Use Aider Safely
A security-first Aider workflow covering Git boundaries, API keys, repository maps, commands, and generated-code review.
Read the guide →DEFEND
AI Security
Threat-aware guidance for prompt injection, agent containment, red teaming, and secure tool use.
Prompt injectionContainmentRed teamingAgent identity
PUBLISHED GUIDES
How to Configure OpenCode Safely
A practical OpenCode security baseline for agents, permissions, shell access, providers, and project configuration.
Read the guide →How to Use GitHub Copilot Agents Safely
A practical security baseline for Copilot coding agent, CLI permissions, repository controls, firewalls, and generated-code review.
Read the guide →How to Configure Roo Code Safely
A practical baseline for Roo Code modes, command execution, MCP servers, secrets, and workspace boundaries.
Read the guide →How to Configure Windsurf Cascade Safely
A security checklist for Windsurf Cascade permissions, terminal actions, workspace rules, secrets, and review gates.
Read the guide →How to Deploy Claude Code Safely: Enterprise Settings, MCP Allowlists, and Defense in Depth
A practical, source-researched blueprint for Claude Code permissions, managed configuration, MCP allowlists, prohibited commands, secret scanning, and EDR.
Read the guide →How to Configure Cline Safely
A security-first Cline setup covering Auto Approve, YOLO mode, terminal commands, browser access, MCP, and checkpoints.
Read the guide →How to Configure Cursor Agent Safely
A security-first Cursor setup covering workspace trust, sandboxing, network access, rules, ignore files, and cloud agents.
Read the guide →How to Configure Gemini CLI Safely
A practical Gemini CLI security baseline covering folder trust, sandboxing, tools, credentials, telemetry, and project settings.
Read the guide →How to Configure OpenAI Codex Safely
A practical security baseline for Codex permissions, sandboxing, network access, project instructions, and approval boundaries.
Read the guide →ON THE RESEARCH DESK
Secure Setup Guides for Codex, Claude Code, and Gemini CLI
Beginner-friendly configuration guides covering permissions, sandboxing, secrets, network access, and review boundaries.
AI Agent Containment: Filesystem, Network, and Credential Boundaries
What strong isolation looks like—and why an approval dialog is not a security boundary.
Codex Security: What the Research Preview Changes
A source-checked look at threat modeling, sandbox validation, and human-reviewed remediation.