Claude Code operates as a command-line interface application that integrates with various development environments. To use it in VS Code, install the beta extension to access a dedicated sidebar panel and inline code suggestions. JetBrains users trigger Claude through the Alt+Enter menu or within the native inspection system. This plugin respects project structures including module boundaries and run configurations. Neovim developers utilize the claudecode.nvim Lua plugin to manage split-pane sessions and visual selection context passing. This plugin allows Claude to edit files through the Neovim buffer system rather than directly on the disk. You can also use the Claude Code SDK to build your own agents or use the Claude Code on GitHub beta by running the /install-github-app command. Claude Code now supports background tasks via GitHub Actions. If you work in the terminal, you can run Claude Code in any editor with an integrated terminal panel.
Plan selection and usage
Developers select plans based on their frequency of coding tasks and the need for administrative control. The Pro plan costs $20 per month or $17 per year and includes terminal access to Claude Code, as well as Claude Cowork, Claude Design, and Claude Docs. The Max 20x plan costs $200 per month and provides 20 times the capacity of the Pro plan for power users who frequently run intensive multi-step coding tasks on large repositories. Organizations using the Team plan pay $20 or $25 per seat depending on the billing cycle, but only Premium seats include Claude Code. Team plans require a minimum of five members and can support up to 150 seats. Enterprise customers pay a $20 per user monthly fee for annual billing, but they must pay for AI usage separately at standard API rates.
| Plan | Monthly Cost | Claude Code Access |
|---|---|---|
| Free | $0 | No |
| Pro | $20 | Yes |
| Max 20x | $200 | Yes |
| Team Premium | $100 (annual) | Yes |
| Enterprise | $20 (annual) + API | Yes |
Usage limits reset weekly and include five-hour session caps. Claude Opus 4.8 costs $5 per million input tokens and $25 per million output tokens via the API. If you set an ANTHROPIC_API_KEY environment variable, Claude Code uses your API credits even if you have a subscription.
Evaluating Model Performance and Security
The market for coding agents changed when GitHub announced Claude Sonnet 4 as the model powering its new coding agent. Claude Opus 5 is the leader for bug fixing, achieving a 96% score on independent SWE-bench Verified evaluations from BenchLM. GitHub Copilot Enterprise and Pro+ users can also select Claude Opus 4.1 from the chat model picker if their administrator enables the policy. Claude Sonnet 4 delivers a 72.7% score on SWE-bench, while Claude Opus 4 achieves 72.5% on SWE-bench and 43.2% on Terminal-bench. Claude Opus 4.6 wins multi-file logic reasoning at 80.8%. Both Claude 4 models are 65% less likely to engage in shortcut or loophole behavior than Sonnet 3.7.
Security management requires constant vigilance because Claude Code operates with the same permissions as the user in the terminal. This permission structure allows the agent to access the local filesystem and execute commands using inherited credentials. Security teams must monitor for unauthorized usage that occurs outside traditional monitoring tools. The MCP Inspector had a remote code execution vulnerability, designated CVE-2025-49596, which had a CVSS score of 9.4. The model can perform reconnaissance or harvest credentials if a user provides those tools. Can an autonomous agent bypass the human-in-the-loop protections required for enterprise compliance?
