Claude Code

What it is

Claude Code is Anthropic’s agentic coding tool: you give it a task in plain language, and it reads the repository, edits files across it, runs shell commands and tests, and commits or opens a pull request. It runs on one engine across several surfaces — a terminal CLI, a VS Code extension (which also installs in Cursor), a JetBrains plugin, the Code tab in the Claude desktop app, the browser at claude.ai/code, and the Claude mobile apps — so a project’s CLAUDE.md, settings and MCP servers follow you between them. It reached the general public in May 2025, and by Anthropic’s own February 2026 funding announcement its run-rate revenue had passed $2.5 billion.

This is a sub-product page. Claude covers the assistant and API; Claude Cowork covers the agent surface for people who do not write code. Claude Code is the one for whoever on the ops team owns code.

Why it shows up in ops stacks

GTM engineers and RevOps developers now ship a growing amount of glue code: CRM sync scripts, enrichment jobs, custom MCP servers, n8n code nodes, warehouse models. Claude Code is where most of that is now built, for four reasons:

• MCP on both sides. It connects to MCP servers for HubSpot, Salesforce, Linear, Notion and Slack, and it is the fastest place to write and test your own server. See what an MCP server is and when a skill beats a server.
• Headless runs. claude -p runs a prompt non-interactively, so it slots into CI, cron jobs and GitHub Actions or GitLab CI/CD pipelines. The Agent SDK exposes the same tools for building your own agents.
• Routines. Cloud-scheduled jobs created with /schedule that keep running when your laptop is off and can fire on an API call or a GitHub event — the overnight data-quality check or the Monday dependency audit nobody has to remember.
• Shared repo instructions. Since version 2.1.277 (18 September 2026), Claude Code reads AGENTS.md when a folder has no CLAUDE.md, so a repository already set up for Codex or other agents works without duplicating instructions. If both files exist, CLAUDE.md wins.

Pricing reality

Claude Code has no price of its own; it is included in every paid Claude plan, and not in Free:

• Pro — $20/mo, or $17/mo billed annually
• Max — $100/mo or $200/mo, at 5× or 20× Pro usage
• Team — standard seat $25/mo ($20 annual); premium seat $125/mo ($100 annual), 2–150 seats
• Enterprise — $20 per seat plus usage billed at API rates

You can also skip seats and run it against an Anthropic Console API key, Amazon Bedrock, Google Cloud’s Agent Platform or Microsoft Foundry, paying per token.

The tier list is not the real number. Usage is pooled across claude.ai chat, the desktop app, Cowork and Claude Code, and is capped twice: a rolling five-hour window and a weekly limit, which Anthropic adjusted several times during 2026. Daily, all-day coding outruns a Pro seat; that is what Max and premium seats are for. Budget a standard Team seat for the analyst who writes a script a week, a premium seat (or Max for individuals) for anyone using it as their main tool, and $100–$200 per heavy user per month as the planning number. Automated GitHub Code Review, available on Team and Enterprise in research preview, is billed separately through usage credits at an average of $15–25 per review.

Best for

GTM engineers, RevOps engineers and legal-ops or recruiting-ops technologists who own a repository — the person who maintains the CRM-to-warehouse sync, writes the custom MCP server for the deal desk, or builds the ATS export pipeline. The scoped case where it wins outright: multi-file changes to an existing codebase that has to keep passing its tests, followed by a scheduled or CI-triggered run of the same job. That is the Team-premium-seat question answered: yes for the one or two people whose job is building that code, no for everyone else.

Not for

People who do not read diffs. Claude Code proposes and runs shell commands and commits code; if nobody on the receiving end can judge whether a change is right, give that person Claude Cowork or a no-code builder instead. It is also not a replacement for a flow tool on high-volume, fixed-shape pipelines: once the job is “move these fields every hour,” the code Claude Code wrote should run in n8n or your orchestrator, not inside an agent session.

Versus the alternatives

• GitHub Copilot — the largest installed base by seat count. Business is $19 and Enterprise $39 per user per month, and since 1 June 2026 usage beyond the included AI credits is metered on top. Pick it when the organization already buys GitHub Enterprise and the main need is inline completion inside the editor for many developers, not delegated multi-step tasks.
• Cursor — the top coding-specific rival by revenue; Bloomberg reported it crossed $2 billion in annualized revenue in March 2026. Individual plans start at $20/mo and Teams at $40 per user. Pick it when the person lives in an IDE and wants to see and steer every edit visually; the full trade-off is in Claude vs Cursor.
• OpenAI Codex — the fastest-growing entrant, from more than 2 million weekly users in March 2026 to more than 5 million by June. It is included in ChatGPT Go ($8), Plus ($20) and Business ($20 per user annual, $25 monthly). Pick it when the company has standardized on ChatGPT Business, so the coding seat comes with no new vendor. OpenAI stopped selling new pay-as-you-go Codex-only seats on Business plans in June 2026, so token-billed use now runs through an API key.

Watch-outs

• Auto mode is now the starting permission mode. From version 2.1.283, interactive terminal and VS Code sessions start in auto mode, where a second model reviews each action instead of a human. Anthropic’s own docs say it “does not guarantee safety.” Guard: on Team and Enterprise, set permissions.defaultMode or set permissions.disableAutoMode to "disable" in managed settings for the groups that touch production systems, add ask rules for any MCP tool that writes to the CRM, and run unattended sessions in a container or sandbox.
• One usage pool, many consumers. A routine, a teammate’s chat session and your coding session draw from the same seat limit, and the limit resets on a five-hour window you do not control. Guard: give scheduled routines and CI jobs their own API key with a spend cap, so a runaway job cannot lock out the person who needs the seat at 10 a.m.
• Code Review cost scales with pushes. Setting a repository to review “after every push” multiplies the $15–25 average by the number of pushes per pull request. Guard: start repositories on “Once after PR creation” or Manual, and set a monthly spend cap for the Claude Code Review service at claude.ai/admin-settings/usage before enabling it.