AI Dev Tools

Installs the agentic coding CLIs — a day-one install class of its own by 2026 — via each vendor's current official channel (most now ship native installers; some are brew/npm), authenticated with subscriptions the user already has. Also the right moment to set security guardrails: these tools get read/write + exec on a machine that holds SSH keys and cloud sessions.

When NOT to use

• Using/configuring agents beyond install (prompts, CLAUDE.md/AGENTS.md content, MCP servers) → project work
• Anthropic/OpenAI API integration in code → SDK docs, not setup
• Local models (Ollama, LM Studio) → different category; mention only
• Editors with AI (Cursor) → editor-setup

Safety rails