Overview
Claude Code is an agentic AI coding assistant developed by Anthropic that enables developers to delegate complex coding tasks from their terminal, IDE, Slack, or web interface. Initially introduced in February 2025 alongside Claude 3.7 Sonnet and reaching general availability in May 2025, it provides deep codebase understanding and can autonomously handle multi-file edits, refactoring, testing, and pull request generation while maintaining human oversight and control.
Designed for professional developers and teams working on real codebases, Claude Code distinguishes itself from traditional autocomplete tools by offering agentic workflows across a project. Anthropic currently lists terminal, web, iOS, Android, GitHub, VS Code, JetBrains, and Slack surfaces. The tool can also connect to external systems through the Model Context Protocol.
Available models and limits depend on the plan and can change faster than a static review. The durable product distinction is the agent workflow and its supported surfaces, not a frozen model roster.
Key Features
Deep Codebase Awareness — Automatically understands your entire project structure, dependencies, and relationships across files without manual context selection, enabling intelligent suggestions and comprehensive refactoring capabilities.
Agentic Task Execution — Handles autonomous coding workflows including feature implementation from descriptions, bug fixing from error messages, test generation, pull request creation, and merge conflict resolution with approval controls.
Multi-Environment Integration — Anthropic lists terminal, web, iOS, Android, GitHub, VS Code, JetBrains, and Slack access, with plan and platform availability subject to change.
Model Context Protocol (MCP) — An open protocol framework that enables connections to external systems via installable MCP servers for platforms like GitHub, Jira, Figma, and Google Drive, with first-class Slack integration built in for team collaboration.
Claude Mods — JavaScript or TypeScript plugins can add interface panels and commands, or handle tool-call events inside Claude Code. Custom UI renders in the terminal and Desktop Code tab; mods run with your permissions and are not sandboxed. See the Mods rollout and upgrade guide.
Model-dependent context — Context and usage limits vary by model, plan, and access method; confirm them in the current product and model documentation for workload planning.
Security & Compliance Controls — Provides data encryption in transit via TLS, permission modes and rules for agent actions, and configurable data retention policies. Enterprise plans add audit logs, admin controls, and custom retention controls. Zero data retention is not included in standard Enterprise; qualified Claude Code Enterprise organizations can have ZDR enabled per organization by their Anthropic account team.
Recent Changes
Claude Mods shipped in v2.1.287 on October 1, 2026. As of October 4, the official latest release is v2.1.289, which adds fixes for mod loading, interface failures, and permission behavior. The Claude Code Mods guide explains installation, supported surfaces, and when existing hooks, skills, or MCP integrations are sufficient.
Pricing & Plans
Claude Code is available across multiple subscription tiers designed for individual developers, teams, and enterprises:
| Plan | Monthly Cost | Key Features | Claude Code Access |
|---|---|---|---|
| Free | $0 | Basic model access via web and mobile apps with strict usage limits | Not included |
| Pro | $20 monthly or approximately $17/month with annual billing ($200 upfront) | Extended usage, model switching capabilities, advanced coding and writing features | Included |
| Max | $100 (5x) / $200 (20x), monthly | 5× or 20× Pro usage per 5-hour session; higher output limits; early feature access; priority at high traffic times | Included |
| Team | Standard: $25/seat monthly or $20/seat/month billed annually; Premium: $125 monthly or $100/seat/month billed annually | For teams of 2–150; Standard and Premium seats; central administration and SSO | Included |
| Enterprise | $20/seat/month, billed annually, plus usage at API rates | Team features plus RBAC, SCIM, audit logs, Compliance API, custom retention, and network controls | Included |
API Pricing: API-billed use is separate from the included Pro or Max subscription allowance. Model prices and caching/batch terms change, so use Anthropic's live API pricing and Claude Code cost documentation for estimates.
Usage limits: Subscription access is not unlimited. Limits depend on plan, model, workload, and current capacity; Anthropic also offers additional usage mechanisms for eligible paid plans.
Best For
- Software engineering teams managing medium to large codebases with frequent cross-file refactoring needs
- Professional developers who prioritize deep code understanding over simple line-level completion
- Organizations with compliance requirements that need audit logs, admin controls, or self-hosted Claude Code execution. Self-hosted environments are a public beta for Team and Enterprise, are unavailable to ZDR organizations, and do not make inference fully private: prompts, responses, and tool results still go to Anthropic.
- Development workflows involving external tool integrations through Model Context Protocol (Slack, Jira, GitHub)
- Teams comfortable with AI-assisted development who maintain rigorous code review practices
Avoid If
- You need only lightweight inline autocomplete and cannot justify an agent-oriented workflow or its usage limits.
- Your organization cannot grant a coding agent scoped access to repositories, terminals, or connected systems under review controls.
- You require a fixed monthly workload guarantee; subscription and API usage limits vary by plan and workload.
Sources & Verification
- Last verified: 2026-10-04
- Evaluation scope: Supported product surfaces, individual and organization pricing, data-use and retention boundaries, and documented integration and execution options; coding quality and cost-per-task were not benchmarked.
- Method: Source-based research from the official product materials below. No hands-on testing was performed for this update.
- Official sources:
FAQ
What models does Claude Code use?
Claude Code uses models made available by Anthropic for the selected plan and access method. Because the roster and defaults change, use the in-product model selector and current documentation rather than a static list here.
Is Claude Code available on mobile devices?
Claude Code is available through web, iOS, and Android in addition to terminal, VS Code, JetBrains, GitHub, and Slack surfaces. Platform availability and plan limits can change, so confirm the current list on Anthropic's product page.
How does Claude Code handle security and data privacy?
Data usage depends on account type. Consumer plans (Free/Pro/Max) allow users to opt out of model training, with 30-day retention when opted out or up to 5 years if opted in. Commercial plans (Team/Enterprise/API) do not use Claude Code prompts or code to train generative models by default, and standard Claude Code retention is 30 days. ZDR is available to qualified Claude Code Enterprise organizations, is not included in standard Enterprise, and is enabled per organization after Anthropic confirms eligibility. Data is encrypted in transit via TLS. Claude Code's permission controls govern agent tool use, with behavior depending on the selected mode. Installed mods run with your user permissions and are not sandboxed; review a mod's source and capabilities before loading it. Enterprise plans add audit logs, admin controls, and compliance APIs for organizational governance.
Can I use Claude Code with my existing IDE?
Yes, Claude Code integrates with Visual Studio Code, including officially documented compatible forks such as Cursor and Devin Desktop, and with JetBrains IDEs through a dedicated plugin. It also works directly from terminal environments including Bash, Zsh, and Fish shells. Additionally, cloud sessions are available on Pro, Max, and Team plans, and for Enterprise users with premium seats or Chat + Claude Code seats, subject to organization controls.
What is the Model Context Protocol?
Model Context Protocol (MCP) is an open framework developed by Anthropic that enables Claude Code to connect with external systems through installable MCP servers. By installing and configuring MCP servers for platforms like GitHub, Jira, Figma, or Google Drive, you can incorporate broader project context into coding workflows and automate cross-platform tasks. Slack is also supported as a first-class integration surface.
Are there usage limits for Claude Code?
Yes. Limits vary by plan, model, workload, and availability. Check Anthropic's current usage documentation and your account meter before planning continuous or high-volume workloads.
How much does Claude Code cost for API usage?
API usage is token-billed separately from subscription allowances. Use Anthropic's live API pricing and Claude Code cost guide with your selected model, caching, and workload profile; this review does not publish a static per-model estimate.
Can Claude Code work with CI/CD pipelines?
Yes, you can integrate Claude Code into CI/CD workflows by scripting the CLI or calling Claude via API in systems like GitHub Actions. This enables automated tasks such as code reviews, test generation, and documentation updates as part of your development pipeline, subject to your organization's security controls. Enterprise and Team plans provide enhanced capabilities for organizational workflow automation.
