[{"data":1,"prerenderedAt":4105},["ShallowReactive",2],{"guide-en-claude-code\u002Fcursor-vs-claude-code":3,"guide-siblings-en-claude-code":721,"guide-alt-en-claude-code\u002Fcursor-vs-claude-code":4104},{"id":4,"title":5,"author":6,"body":7,"date":693,"description":694,"extension":695,"faq":696,"meta":712,"navigation":713,"order":134,"path":714,"readTime":715,"seo":716,"stem":717,"topic":718,"translationId":719,"updated":693,"__hash__":720},"guides\u002Fguides\u002Fclaude-code\u002Fcursor-vs-claude-code.md","Cursor vs Claude Code: which AI coding tool fits your team?","Walma Engineering",{"type":8,"value":9,"toc":682},"minimark",[10,20,25,42,45,49,58,92,99,104,138,150,153,157,225,228,231,235,238,242,274,278,314,344,347,351,354,425,433,437,440,468,521,546,565,569,579,596,599,603,649,653,665,678],[11,12,13,14,19],"p",{},"\"Cursor vs Claude Code\" is one of the most searched questions in AI coding right now, and most answers are out of date within a quarter. Both tools have converged a lot: Cursor now has a terminal agent and cloud agents, Claude Code now has IDE extensions and a desktop app. The difference that remains is where each one starts from. This guide compares them as of October 2026, based on the vendors' own docs and pricing pages. If you are new to Claude Code, read the ",[15,16,18],"a",{"href":17},"\u002Fen\u002Fguides\u002Fclaude-code","overview"," first.",[21,22,24],"h2",{"id":23},"the-short-version","The short version",[26,27,28,36],"ul",{},[29,30,31,35],"li",{},[32,33,34],"strong",{},"Cursor"," is an editor with AI built into every surface: Tab autocomplete as you type, an agent panel, cloud agents, and a choice of models from several labs.",[29,37,38,41],{},[32,39,40],{},"Claude Code"," is an agent with several front ends: the CLI, VS Code and JetBrains extensions, a desktop app, the web, Slack and CI. It runs Claude models only.",[11,43,44],{},"If your developers live in the editor and want fast inline help with occasional agent tasks, Cursor fits more naturally. If you want an agent that does long multi-step work, scripts into CI and behaves the same across every surface, Claude Code fits better. Many teams end up using both, and that works fine (see the last section).",[21,46,48],{"id":47},"interface-ide-first-vs-agent-first","Interface: IDE first vs agent first",[11,50,51,53,54,57],{},[32,52,34],{}," is a fork of VS Code. You get the familiar editor, extensions and keybindings, with AI layered on top. ",[32,55,56],{},"Tab"," is Cursor's autocomplete, which in Cursor's words \"suggests code as you type, based on your recent edits, surrounding code, and linter errors\". The agent works in a side panel and shows diffs in the editor. Cursor also ships a CLI, installed with:",[59,60,65],"pre",{"className":61,"code":62,"language":63,"meta":64,"style":64},"language-bash shiki shiki-themes github-dark","curl https:\u002F\u002Fcursor.com\u002Finstall -fsS | bash\n","bash","",[66,67,68],"code",{"__ignoreMap":64},[69,70,73,77,81,85,89],"span",{"class":71,"line":72},"line",1,[69,74,76],{"class":75},"svObZ","curl",[69,78,80],{"class":79},"sU2Wk"," https:\u002F\u002Fcursor.com\u002Finstall",[69,82,84],{"class":83},"sDLfK"," -fsS",[69,86,88],{"class":87},"snl16"," |",[69,90,91],{"class":75}," bash\n",[11,93,94,95,98],{},"The binary is called ",[66,96,97],{},"agent",", and it has a print mode for scripts and CI.",[11,100,101,103],{},[32,102,40],{}," started as a terminal tool, and the CLI is still the most complete surface:",[59,105,107],{"className":61,"code":106,"language":63,"meta":64,"style":64},"curl -fsSL https:\u002F\u002Fclaude.ai\u002Finstall.sh | bash\ncd your-project\nclaude\n",[66,108,109,123,132],{"__ignoreMap":64},[69,110,111,113,116,119,121],{"class":71,"line":72},[69,112,76],{"class":75},[69,114,115],{"class":83}," -fsSL",[69,117,118],{"class":79}," https:\u002F\u002Fclaude.ai\u002Finstall.sh",[69,120,88],{"class":87},[69,122,91],{"class":75},[69,124,126,129],{"class":71,"line":125},2,[69,127,128],{"class":83},"cd",[69,130,131],{"class":79}," your-project\n",[69,133,135],{"class":71,"line":134},3,[69,136,137],{"class":75},"claude\n",[11,139,140,141,144,145,149],{},"Around that engine Anthropic now ships a VS Code extension (inline diffs, @-mentions, plan review, conversation history), a JetBrains plugin, a desktop app where you run sessions side by side and review diffs visually, and Claude Code on the web at claude.ai\u002Fcode. Anthropic's docs say every surface \"connects to the same underlying Claude Code engine\", so a repo's ",[66,142,143],{},"CLAUDE.md",", settings and MCP servers carry across. Installation details are in ",[15,146,148],{"href":147},"\u002Fen\u002Fguides\u002Fclaude-code\u002Fhow-to-install-claude-code","how to install Claude Code",".",[11,151,152],{},"The practical gap: Claude Code has no Tab-style autocomplete. If inline completion is a big part of how your team writes code, that matters.",[21,154,156],{"id":155},"models","Models",[158,159,160,173],"table",{},[161,162,163],"thead",{},[164,165,166,169,171],"tr",{},[167,168],"th",{},[167,170,34],{},[167,172,40],{},[174,175,176,188,199,214],"tbody",{},[164,177,178,182,185],{},[179,180,181],"td",{},"Model providers",[179,183,184],{},"Anthropic, OpenAI, Google and others",[179,186,187],{},"Anthropic only",[164,189,190,193,196],{},[179,191,192],{},"Own model",[179,194,195],{},"Composer (currently 2.5)",[179,197,198],{},"No",[164,200,201,204,207],{},[179,202,203],{},"Automatic model choice",[179,205,206],{},"Auto, with Cost, Balance and Intelligence modes",[179,208,209,210,213],{},"Aliases such as ",[66,211,212],{},"opusplan"," (Opus for planning, Sonnet for execution)",[164,215,216,219,222],{},[179,217,218],{},"Hosting options",[179,220,221],{},"Cursor's service",[179,223,224],{},"Anthropic API, Amazon Bedrock, Google Cloud, Microsoft Foundry, LLM gateways",[11,226,227],{},"Cursor's strength is choice: you can run the same task on Claude, GPT or Gemini and keep whichever works best. Claude Code's strength is depth: the harness is tuned for one model family by the people who train it, and you choose where that model is hosted. For EU companies that last point is often decisive, because it lets you keep inference in an EU region.",[11,229,230],{},"One billing detail people miss: when you use Claude inside Cursor, it is billed against your Cursor plan at the model's API rate. Your Claude subscription does not apply there.",[21,232,234],{"id":233},"agent-capabilities","Agent capabilities",[11,236,237],{},"Both tools now do the core agent loop well: read the codebase, plan, edit many files, run commands and tests, fix what fails, and commit. The differences are in what surrounds that loop.",[11,239,240],{},[32,241,34],{},[26,243,244,250,256],{},[29,245,246,249],{},[32,247,248],{},"Cloud agents"," run in isolated cloud VMs with a full development environment. You can start them from the desktop app, cursor.com\u002Fagents, the iOS app, Slack, GitHub or Bitbucket comments, Linear and an API.",[29,251,252,255],{},[32,253,254],{},"Bugbot"," does agentic code review on pull requests.",[29,257,258,261,262,265,266,269,270,273],{},[32,259,260],{},"Hooks"," and ",[32,263,264],{},"rules"," (",[66,267,268],{},".cursor\u002Frules",", ",[66,271,272],{},"AGENTS.md",") shape agent behaviour.",[11,275,276],{},[32,277,40],{},[26,279,280,286,295,305],{},[29,281,282,285],{},[32,283,284],{},"Subagents"," split work across several agents, with a lead agent coordinating.",[29,287,288,265,291,294],{},[32,289,290],{},"Headless mode",[66,292,293],{},"claude -p",") pipes into scripts, and there are official GitHub Actions and GitLab CI\u002FCD integrations.",[29,296,297,300,301,304],{},[32,298,299],{},"Routines"," run on a schedule in the cloud; ",[66,302,303],{},"\u002Floop"," repeats a prompt in a session.",[29,306,307,309,310,313],{},[32,308,260],{}," run shell commands before or after actions, and the ",[32,311,312],{},"Agent SDK"," lets you build your own agents on the same tooling.",[59,315,317],{"className":61,"code":316,"language":63,"meta":64,"style":64},"git diff main --name-only | claude -p \"review these changed files for security issues\"\n",[66,318,319],{"__ignoreMap":64},[69,320,321,324,327,330,333,335,338,341],{"class":71,"line":72},[69,322,323],{"class":75},"git",[69,325,326],{"class":79}," diff",[69,328,329],{"class":79}," main",[69,331,332],{"class":83}," --name-only",[69,334,88],{"class":87},[69,336,337],{"class":75}," claude",[69,339,340],{"class":83}," -p",[69,342,343],{"class":79}," \"review these changed files for security issues\"\n",[11,345,346],{},"Our take: Claude Code is the more programmable of the two. Cursor's cloud agents are the more polished way to hand a ticket to an agent from Slack or Linear and get a branch back.",[21,348,350],{"id":349},"pricing","Pricing",[11,352,353],{},"Prices from the vendors' pages, in US dollars per month, as of October 2026.",[158,355,356,368],{},[161,357,358],{},[164,359,360,363,365],{},[167,361,362],{},"Tier",[167,364,34],{},[167,366,367],{},"Claude (includes Claude Code)",[174,369,370,381,392,403,414],{},[164,371,372,375,378],{},[179,373,374],{},"Free",[179,376,377],{},"Hobby: limited agent requests",[179,379,380],{},"Free plan: Claude Code not included",[164,382,383,386,389],{},[179,384,385],{},"Entry",[179,387,388],{},"Pro: $20",[179,390,391],{},"Pro: $20 monthly, $17 billed annually",[164,393,394,397,400],{},[179,395,396],{},"Heavy individual",[179,398,399],{},"Pro+: $60, Ultra: $200",[179,401,402],{},"Max: from $100 (5x or 20x Pro usage)",[164,404,405,408,411],{},[179,406,407],{},"Team",[179,409,410],{},"Teams Standard: $40\u002Fuser, Teams Premium: $120\u002Fuser",[179,412,413],{},"Team Standard: $25\u002Fseat ($20 annual), Team Premium: $125\u002Fseat ($100 annual)",[164,415,416,419,422],{},[179,417,418],{},"Enterprise",[179,420,421],{},"Custom",[179,423,424],{},"$20\u002Fseat billed annually, plus usage at API rates",[11,426,427,428,432],{},"Both vendors meter usage, not just seats. Cursor says model choice \"affects how quickly your included usage is consumed\", and past the limit you continue at the same API rates. Claude plans have usage windows, and Claude Code can also be billed per token through the API. Our ",[15,429,431],{"href":430},"\u002Fen\u002Fguides\u002Fclaude-code\u002Fclaude-code-pricing","Claude Code pricing guide"," goes through those options in detail.",[21,434,436],{"id":435},"mcp-skills-and-rules","MCP, skills and rules",[11,438,439],{},"This is where the two have converged most.",[11,441,442,445,446,449,450,453,454,457,458,461,462,261,465,149],{},[32,443,444],{},"MCP."," Both are full MCP clients with local and remote servers. Cursor reads ",[66,447,448],{},".cursor\u002Fmcp.json"," per project and ",[66,451,452],{},"~\u002F.cursor\u002Fmcp.json"," globally; Claude Code uses ",[66,455,456],{},"claude mcp add"," and a shared ",[66,459,460],{},".mcp.json",". Setup guides: ",[15,463,34],{"href":464},"\u002Fen\u002Fguides\u002Fmcp\u002Fadd-mcp-server-cursor",[15,466,40],{"href":467},"\u002Fen\u002Fguides\u002Fmcp\u002Fadd-mcp-server-claude-code",[59,469,473],{"className":470,"code":471,"language":472,"meta":64,"style":64},"language-json shiki shiki-themes github-dark","{\n  \"mcpServers\": {\n    \"notion\": { \"url\": \"https:\u002F\u002Fmcp.notion.com\u002Fmcp\" }\n  }\n}\n","json",[66,474,475,481,489,509,515],{"__ignoreMap":64},[69,476,477],{"class":71,"line":72},[69,478,480],{"class":479},"s95oV","{\n",[69,482,483,486],{"class":71,"line":125},[69,484,485],{"class":83},"  \"mcpServers\"",[69,487,488],{"class":479},": {\n",[69,490,491,494,497,500,503,506],{"class":71,"line":134},[69,492,493],{"class":83},"    \"notion\"",[69,495,496],{"class":479},": { ",[69,498,499],{"class":83},"\"url\"",[69,501,502],{"class":479},": ",[69,504,505],{"class":79},"\"https:\u002F\u002Fmcp.notion.com\u002Fmcp\"",[69,507,508],{"class":479}," }\n",[69,510,512],{"class":71,"line":511},4,[69,513,514],{"class":479},"  }\n",[69,516,518],{"class":71,"line":517},5,[69,519,520],{"class":479},"}\n",[11,522,523,526,527,530,531,261,534,537,538,261,542,149],{},[32,524,525],{},"Skills."," Both support Agent Skills, folders with a ",[66,528,529],{},"SKILL.md",". Cursor loads its own directories and, for compatibility, also ",[66,532,533],{},".claude\u002Fskills\u002F",[66,535,536],{},"~\u002F.claude\u002Fskills\u002F",". That means one skills folder in the repo can serve both tools. See ",[15,539,541],{"href":540},"\u002Fen\u002Fguides\u002Fskills","what skills are",[15,543,545],{"href":544},"\u002Fen\u002Fguides\u002Fskills\u002Fclaude-skills-vs-mcp","skills vs MCP",[11,547,548,551,552,554,555,557,558,261,560,562,563,149],{},[32,549,550],{},"Instructions."," Claude Code reads ",[66,553,143],{}," and can also read ",[66,556,272],{},". Cursor reads ",[66,559,268],{},[66,561,272],{},". If you use both tools, keep shared conventions in ",[66,564,272],{},[21,566,568],{"id":567},"team-and-enterprise-controls","Team and enterprise controls",[11,570,571,574,575,578],{},[32,572,573],{},"Cursor Teams"," adds central billing and administration, team-wide privacy mode, SAML\u002FOIDC SSO, usage analytics and a team marketplace for rules, skills and plugins. ",[32,576,577],{},"Cursor Enterprise"," adds pooled usage, SCIM, repository, model and MCP access controls, auto-run, browser and network controls, and audit logs.",[11,580,581,583,584,587,588,591,592,595],{},[32,582,40],{}," is governed through managed settings: a ",[66,585,586],{},"managed-settings.json"," file, an MDM policy or server-managed settings from the claude.ai console. Anthropic's docs say \"nothing you set overrides them\". Admins can lock permissions (",[66,589,590],{},"allowManagedPermissionRulesOnly",") and restrict models (",[66,593,594],{},"availableModels","). Claude Enterprise adds user and org spend limits and role-based access.",[11,597,598],{},"Both are serious about control at the top tier. The difference is scope: each governs its own tool. Neither governs the other, the chat assistant in the browser, or the agents your engineers build with the API.",[21,600,602],{"id":601},"when-to-pick-which","When to pick which",[158,604,605,615],{},[161,606,607],{},[164,608,609,612],{},[167,610,611],{},"Pick Cursor if",[167,613,614],{},"Pick Claude Code if",[174,616,617,625,633,641],{},[164,618,619,622],{},[179,620,621],{},"Inline autocomplete is central to how people code",[179,623,624],{},"Most work is multi-step agent tasks",[164,626,627,630],{},[179,628,629],{},"You want to switch between models from several labs",[179,631,632],{},"You want Claude, hosted where you choose",[164,634,635,638],{},[179,636,637],{},"Your team is happy in a VS Code style editor",[179,639,640],{},"Your team uses JetBrains, Vim or mixed editors",[164,642,643,646],{},[179,644,645],{},"You want agents launched from Slack, Linear or GitHub with little setup",[179,647,648],{},"You want to script the agent into CI, cron and your own tooling",[21,650,652],{"id":651},"using-both","Using both",[11,654,655,656,659,660,269,662,664],{},"Running both is common and low friction. Anthropic publishes an install link for the Claude Code extension in Cursor, and the ",[66,657,658],{},"claude"," CLI works in Cursor's integrated terminal. A typical split: Cursor's Tab and quick edits for the moment-to-moment work, Claude Code for larger refactors, test backfills and CI jobs. Keep ",[66,661,272],{},[66,663,533],{}," and the project MCP config in the repo so both tools read the same conventions.",[11,666,667,668,672,673,677],{},"The cost is governance. Two tools mean two admin consoles, two sets of MCP permissions, two bills and two places where code and prompts leave your network. That is the problem we built Walma AI Hub for: a gateway in your own Azure tenant in an EU region that Claude Code, Cursor and your other AI tools route through, with one model policy, one MCP allowlist, one budget and one log. If that is the question on your desk, the ",[15,669,671],{"href":670},"\u002Fen\u002Fai-hub","AI Hub page"," has the details, and ",[15,674,676],{"href":675},"\u002Fen\u002Fguides\u002Fmcp\u002Fmcp-gateway","MCP gateways"," explains the pattern.",[679,680,681],"style",{},"html pre.shiki code .svObZ, html code.shiki .svObZ{--shiki-default:#B392F0}html pre.shiki code .sU2Wk, html code.shiki .sU2Wk{--shiki-default:#9ECBFF}html pre.shiki code .sDLfK, html code.shiki .sDLfK{--shiki-default:#79B8FF}html pre.shiki code .snl16, html code.shiki .snl16{--shiki-default:#F97583}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html pre.shiki code .s95oV, html code.shiki .s95oV{--shiki-default:#E1E4E8}",{"title":64,"searchDepth":125,"depth":134,"links":683},[684,685,686,687,688,689,690,691,692],{"id":23,"depth":125,"text":24},{"id":47,"depth":125,"text":48},{"id":155,"depth":125,"text":156},{"id":233,"depth":125,"text":234},{"id":349,"depth":125,"text":350},{"id":435,"depth":125,"text":436},{"id":567,"depth":125,"text":568},{"id":601,"depth":125,"text":602},{"id":651,"depth":125,"text":652},"2026-10-05","An honest comparison of Cursor and Claude Code as of October 2026: interface, models, agent capabilities, pricing tiers, MCP and skills support, team and enterprise controls, when to pick which, and how to run both side by side.","md",[697,700,703,706,709],{"q":698,"a":699},"What is the main difference between Cursor and Claude Code?","Cursor is an AI code editor: a VS Code based IDE with autocomplete, an agent panel and cloud agents built in. Claude Code is Anthropic's coding agent: it started in the terminal and now also runs as a VS Code and JetBrains extension, a desktop app and on the web. Cursor is built around the editor, Claude Code around the agent.",{"q":701,"a":702},"Can I use Claude models in Cursor?","Yes. Cursor offers models from Anthropic, OpenAI, Google and others, plus its own Composer model. Claude usage in Cursor is billed against your Cursor plan at the model's API rate, not against a Claude subscription.",{"q":704,"a":705},"Can I use GPT or Gemini in Claude Code?","No. Claude Code runs Anthropic's Claude models only. You can reach those models through Anthropic directly or through Amazon Bedrock, Google Cloud, Microsoft Foundry or an LLM gateway, but the models themselves are always Claude.",{"q":707,"a":708},"Can I run Claude Code inside Cursor?","Yes. Anthropic's docs list an install link for Cursor, and the Claude Code extension is also on the Open VSX registry. You can also just run the claude CLI in Cursor's integrated terminal.",{"q":710,"a":711},"Which is cheaper, Cursor or Claude Code?","Both start at 20 US dollars a month for an individual plan (Cursor Pro, Claude Pro). Heavy users move up to Cursor Pro+ (60) or Ultra (200), or to Claude Max (from 100). For teams, Cursor Teams is 40 per user per month and Claude Team starts at 25 per seat billed monthly. The real cost depends on how much agent work each person runs.",{},true,"\u002Fguides\u002Fclaude-code\u002Fcursor-vs-claude-code","8 min read",{"title":5,"description":694},"guides\u002Fclaude-code\u002Fcursor-vs-claude-code","claude-code",null,"yAo9uj_30aqYbxTadAGSbIsxAraaW1zmp-0-P77WpMM",[722,1078,1560,1808,2270,3148],{"id":723,"title":724,"author":6,"body":725,"date":1052,"description":1053,"extension":695,"faq":1054,"meta":1070,"navigation":713,"order":1071,"path":1072,"readTime":1073,"seo":1074,"stem":1075,"topic":718,"translationId":1076,"updated":1052,"__hash__":1077},"guides\u002Fguides\u002Fclaude-code\u002Findex.md","What is Claude Code? A complete guide to Anthropic's coding agent",{"type":8,"value":726,"toc":1043},[727,730,733,737,740,781,784,788,791,837,840,859,868,872,875,901,907,911,921,927,944,955,961,965,971,977,983,987,990,996,1002,1008,1015,1019,1040],[11,728,729],{},"Claude Code is a tool you start in your terminal and give tasks in plain language: \"Find out why the login test fails and fix it.\" The agent reads the codebase, plans, changes files, runs the tests and shows you the result. Since its launch in early 2025 it has become one of the most widely used coding agents, and for many companies the first place they meet agentic AI in production.",[11,731,732],{},"This is the starting point. It covers what Claude Code can do, how it compares with the alternatives, the key concepts, and what a company needs to know about data and governance. Installation and pricing have their own pages.",[21,734,736],{"id":735},"what-claude-code-does","What Claude Code does",[11,738,739],{},"Unlike autocomplete in an editor, Claude Code works as an agent: it takes a goal and executes several steps on its own.",[26,741,742,748,754,760,766,772],{},[29,743,744,747],{},[32,745,746],{},"Understands the codebase."," Searches the project, reads relevant files, explains architecture or individual functions.",[29,749,750,753],{},[32,751,752],{},"Makes changes."," New features, refactors, bug fixes across many files, with a diff preview before every edit.",[29,755,756,759],{},[32,757,758],{},"Runs and verifies."," Tests, linters, build commands. If something fails it reads the output and corrects.",[29,761,762,765],{},[32,763,764],{},"Handles git."," Branches, commits with sensible messages, pull requests, merge conflicts.",[29,767,768,771],{},[32,769,770],{},"Connects to tools."," Through MCP servers it reaches GitHub, Jira, databases, browsers and internal systems.",[29,773,774,777,778,780],{},[32,775,776],{},"Loads know-how."," Through skills and a ",[66,779,143],{}," file in the repository it knows your team's conventions.",[11,782,783],{},"It runs in the terminal, as an extension in VS Code and JetBrains, as a GitHub Action in CI, and through an SDK inside your own agents.",[21,785,787],{"id":786},"installation-in-brief","Installation in brief",[11,789,790],{},"The fastest route is the native installer, which needs no Node.js:",[59,792,794],{"className":61,"code":793,"language":63,"meta":64,"style":64},"# macOS and Linux\ncurl -fsSL https:\u002F\u002Fclaude.ai\u002Finstall.sh | bash\n\n# Windows (PowerShell)\nirm https:\u002F\u002Fclaude.ai\u002Finstall.ps1 | iex\n",[66,795,796,802,814,819,824],{"__ignoreMap":64},[69,797,798],{"class":71,"line":72},[69,799,801],{"class":800},"sAwPA","# macOS and Linux\n",[69,803,804,806,808,810,812],{"class":71,"line":125},[69,805,76],{"class":75},[69,807,115],{"class":83},[69,809,118],{"class":79},[69,811,88],{"class":87},[69,813,91],{"class":75},[69,815,816],{"class":71,"line":134},[69,817,818],{"emptyLinePlaceholder":713},"\n",[69,820,821],{"class":71,"line":511},[69,822,823],{"class":800},"# Windows (PowerShell)\n",[69,825,826,829,832,834],{"class":71,"line":517},[69,827,828],{"class":75},"irm",[69,830,831],{"class":79}," https:\u002F\u002Fclaude.ai\u002Finstall.ps1",[69,833,88],{"class":87},[69,835,836],{"class":75}," iex\n",[11,838,839],{},"Or via npm if Node.js 18 or later is present:",[59,841,843],{"className":61,"code":842,"language":63,"meta":64,"style":64},"npm install -g @anthropic-ai\u002Fclaude-code\n",[66,844,845],{"__ignoreMap":64},[69,846,847,850,853,856],{"class":71,"line":72},[69,848,849],{"class":75},"npm",[69,851,852],{"class":79}," install",[69,854,855],{"class":83}," -g",[69,857,858],{"class":79}," @anthropic-ai\u002Fclaude-code\n",[11,860,861,862,864,865,149],{},"Then run ",[66,863,658],{}," in a project folder and sign in on first launch. Step-by-step instructions, common errors and enterprise setups are in ",[15,866,867],{"href":147},"How to install Claude Code",[21,869,871],{"id":870},"what-claude-code-costs","What Claude Code costs",[11,873,874],{},"Claude Code is not sold separately. It is included in Anthropic's subscriptions or billed per token through the API:",[26,876,877,883,889,895],{},[29,878,879,882],{},[32,880,881],{},"Pro",": the entry point for individuals, with a usage allowance.",[29,884,885,888],{},[32,886,887],{},"Max",": much higher allowances in two tiers, for developers who use the agent all day.",[29,890,891,894],{},[32,892,893],{},"Team and Enterprise",": seats with central administration; Claude Code is in the premium seats.",[29,896,897,900],{},[32,898,899],{},"API",": per-token billing, right for CI pipelines and custom agents.",[11,902,903,904,149],{},"The amounts, what \"allowance\" means in practice, and how to control cost across a team are in ",[15,905,906],{"href":430},"Is Claude Code free? Pricing explained",[21,908,910],{"id":909},"the-key-concepts","The key concepts",[11,912,913,916,917,920],{},[32,914,915],{},"CLAUDE.md."," A Markdown file in the repository, loaded at the start of every session. Build and test commands, conventions, and anything the agent must always know. ",[66,918,919],{},"\u002Finit"," drafts one from what it finds in the repo.",[11,922,923,926],{},[32,924,925],{},"Permissions."," Claude Code asks before it edits files or runs commands. You can allow specific commands permanently, keep the agent in a read-only plan mode, or let it run more freely in trusted environments such as a sandbox or CI.",[11,928,929,932,933,935,936,939,940,149],{},[32,930,931],{},"MCP servers."," Through the Model Context Protocol the agent gets access to external systems. Add one with ",[66,934,456],{},"; inside a session ",[66,937,938],{},"\u002Fmcp"," shows status and starts authentication. See ",[15,941,943],{"href":942},"\u002Fen\u002Fguides\u002Fmcp\u002Fwhat-is-an-mcp-server","What is an MCP server?",[11,945,946,948,949,951,952,149],{},[32,947,525],{}," Folders with a ",[66,950,529],{}," that teach the agent how a specific task is done in your team, loaded only when relevant. See ",[15,953,954],{"href":540},"Claude skills explained",[11,956,957,960],{},[32,958,959],{},"Hooks, subagents, plugins."," Hooks run your own scripts before or after certain actions (a formatter after every edit, say). Subagents handle subtasks in their own context. Plugins bundle skills, commands, hooks and MCP servers for distribution to a team.",[21,962,964],{"id":963},"claude-code-compared","Claude Code compared",[11,966,967,970],{},[32,968,969],{},"Claude Code vs Cursor."," Cursor is an editor with AI built in. Claude Code is an agent that plugs into any editor and into CI. Cursor is stronger for interactive editing, Claude Code for longer multi-step tasks. Many teams use both, often with Claude models in both.",[11,972,973,976],{},[32,974,975],{},"Claude Code vs Codex."," OpenAI's Codex is the direct counterpart with GPT models: CLI, editor integration, cloud execution. The choice usually comes down to which models the company has approved. A gateway that serves both takes the heat out of the decision.",[11,978,979,982],{},[32,980,981],{},"Claude Code vs GitHub Copilot."," Copilot started as autocomplete and now has an agent mode with model choice, including Claude. It is tightly tied to GitHub. Claude Code is more ecosystem-independent and more deeply built for agentic work.",[21,984,986],{"id":985},"data-protection-and-running-it-in-the-eu","Data protection and running it in the EU",[11,988,989],{},"This is where most European rollouts stall, and it is solvable.",[11,991,992,995],{},[32,993,994],{},"Default operation."," With an Anthropic account, requests go through Anthropic's API in the US. Team, Enterprise and API usage fall under the commercial terms, under which data is not used for training, and a data processing agreement is available.",[11,997,998,1001],{},[32,999,1000],{},"EU regions."," Claude Code supports Amazon Bedrock, Google Vertex AI and Microsoft Foundry as backends, so model calls can run in an EU region such as Frankfurt or Sweden Central. The switch is a few environment variables; developers keep the same tool.",[11,1003,1004,1007],{},[32,1005,1006],{},"What a gateway adds."," Even with an EU region, questions remain: who may use which models, with what budget? Which MCP servers are approved? Where do credentials live? What did the agent do yesterday at 14:32? A gateway between developer and model answers all of them: policy, budgets, approved servers and a full log in one place. Claude Code supports a configurable base URL, so every request can go through the gateway without the developer noticing.",[11,1009,1010,1011,1014],{},"Walma AI Hub is that gateway, running in the customer's own Azure tenant in an EU region, for Claude Code, Codex, Cursor and the MCP servers the company approves. Developers install one signed client and work with one key; there is no client-side switch to turn policy off. If you are rolling Claude Code out beyond a handful of developers, a ",[15,1012,1013],{"href":670},"20-minute walkthrough"," is the fastest way to settle the open questions.",[21,1016,1018],{"id":1017},"further-reading","Further reading",[26,1020,1021,1026,1030,1035],{},[29,1022,1023,1025],{},[15,1024,867],{"href":147}," on macOS, Linux and Windows.",[29,1027,1028,149],{},[15,1029,906],{"href":430},[29,1031,1032,1034],{},[15,1033,943],{"href":942}," How the agent reaches tools and data.",[29,1036,1037,1039],{},[15,1038,954],{"href":540},". Making team know-how reusable.",[679,1041,1042],{},"html pre.shiki code .sAwPA, html code.shiki .sAwPA{--shiki-default:#6A737D}html pre.shiki code .svObZ, html code.shiki .svObZ{--shiki-default:#B392F0}html pre.shiki code .sDLfK, html code.shiki .sDLfK{--shiki-default:#79B8FF}html pre.shiki code .sU2Wk, html code.shiki .sU2Wk{--shiki-default:#9ECBFF}html pre.shiki code .snl16, html code.shiki .snl16{--shiki-default:#F97583}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}",{"title":64,"searchDepth":125,"depth":134,"links":1044},[1045,1046,1047,1048,1049,1050,1051],{"id":735,"depth":125,"text":736},{"id":786,"depth":125,"text":787},{"id":870,"depth":125,"text":871},{"id":909,"depth":125,"text":910},{"id":963,"depth":125,"text":964},{"id":985,"depth":125,"text":986},{"id":1017,"depth":125,"text":1018},"2026-09-11","Claude Code is Anthropic's agentic coding tool that runs in your terminal, reads your codebase, edits files, runs tests and opens pull requests. This guide covers what it does, how it compares with Cursor, Codex and Copilot, how to install it, what it costs, and how to run it in an EU environment with policy and logging.",[1055,1058,1061,1064,1067],{"q":1056,"a":1057},"What is Claude Code?","Claude Code is a command-line tool from Anthropic that runs Claude as a coding agent in your terminal. It reads your codebase, plans changes, edits files, runs tests and commands, and creates commits and pull requests, in a conversation with you.",{"q":1059,"a":1060},"Is Claude Code free?","No. Claude Code is included in the Pro, Max, Team and Enterprise plans, or billed per token through the API. There is no permanently free tier, though new API accounts sometimes get trial credits.",{"q":1062,"a":1063},"Does Claude Code work on Windows?","Yes. Claude Code runs natively on Windows 10 and 11 (with Git for Windows), on macOS and Linux, and inside WSL. Install it with the native installer script or via npm.",{"q":1065,"a":1066},"Claude Code vs Cursor: which is better?","They are different tools. Cursor is an editor with an AI assistant built in; Claude Code is an agent that works from the terminal, independent of your editor, and can be embedded in VS Code, JetBrains and CI. Cursor is stronger for interactive editing, Claude Code for longer multi-step tasks. Many teams use both.",{"q":1068,"a":1069},"Where is my code processed when I use Claude Code?","By default at Anthropic in the US. Companies in the EU can run Claude Code through Amazon Bedrock, Google Vertex AI or Microsoft Foundry in an EU region, or through a gateway such as Walma AI Hub in their own Azure tenant.",{},0,"\u002Fguides\u002Fclaude-code","12 min read",{"title":724,"description":1053},"guides\u002Fclaude-code\u002Findex","claude-code-guide","_2hCTbI-2-EcxQYApmXsrcjnQhUGQJgYBYjJC9880eM",{"id":1079,"title":1080,"author":6,"body":1081,"date":1052,"description":1537,"extension":695,"faq":1538,"meta":1554,"navigation":713,"order":72,"path":1555,"readTime":715,"seo":1556,"stem":1557,"topic":718,"translationId":1558,"updated":1052,"__hash__":1559},"guides\u002Fguides\u002Fclaude-code\u002Fhow-to-install-claude-code.md","How to install Claude Code on macOS, Linux and Windows",{"type":8,"value":1082,"toc":1526},[1083,1091,1095,1131,1134,1138,1141,1162,1196,1199,1211,1215,1218,1232,1239,1243,1246,1262,1265,1268,1287,1291,1297,1305,1308,1311,1374,1378,1381,1401,1405,1417,1423,1433,1442,1446,1449,1478,1481,1485,1488,1517,1523],[11,1084,1085,1086,1088,1089,149],{},"Installing Claude Code takes under two minutes. The problems come afterwards: not signed in, wrong project folder, no ",[66,1087,143],{},", or the question of where the data actually goes. This guide walks through everything in order. For what Claude Code is and does, start with the ",[15,1090,18],{"href":17},[21,1092,1094],{"id":1093},"requirements","Requirements",[26,1096,1097,1111,1118,1121],{},[29,1098,1099,1102,1103,1106,1107,1110],{},[32,1100,1101],{},"macOS"," 10.15 or later, ",[32,1104,1105],{},"Linux"," (Ubuntu 20.04+, Debian 10+ or equivalent) or ",[32,1108,1109],{},"Windows"," 10\u002F11.",[29,1112,1113,1114,1117],{},"On Windows: ",[32,1115,1116],{},"Git for Windows",", because Claude Code uses its Bash for shell commands. WSL 1 or 2 also works.",[29,1119,1120],{},"A terminal and internet access.",[29,1122,1123,1124,1127,1128,1130],{},"An Anthropic account (Pro, Max, Team or Enterprise) ",[32,1125,1126],{},"or"," an API key ",[32,1129,1126],{}," access to Bedrock, Vertex AI or Microsoft Foundry.",[11,1132,1133],{},"Node.js is only needed for the npm route.",[21,1135,1137],{"id":1136},"option-1-native-installer-recommended","Option 1: Native installer (recommended)",[11,1139,1140],{},"The native installer places a self-contained binary and keeps it up to date automatically.",[59,1142,1144],{"className":61,"code":1143,"language":63,"meta":64,"style":64},"# macOS and Linux\ncurl -fsSL https:\u002F\u002Fclaude.ai\u002Finstall.sh | bash\n",[66,1145,1146,1150],{"__ignoreMap":64},[69,1147,1148],{"class":71,"line":72},[69,1149,801],{"class":800},[69,1151,1152,1154,1156,1158,1160],{"class":71,"line":125},[69,1153,76],{"class":75},[69,1155,115],{"class":83},[69,1157,118],{"class":79},[69,1159,88],{"class":87},[69,1161,91],{"class":75},[59,1163,1167],{"className":1164,"code":1165,"language":1166,"meta":64,"style":64},"language-powershell shiki shiki-themes github-dark","# Windows, PowerShell\nirm https:\u002F\u002Fclaude.ai\u002Finstall.ps1 | iex\n","powershell",[66,1168,1169,1174],{"__ignoreMap":64},[69,1170,1171],{"class":71,"line":72},[69,1172,1173],{"class":800},"# Windows, PowerShell\n",[69,1175,1176,1179,1182,1185,1188,1191,1194],{"class":71,"line":125},[69,1177,1178],{"class":479},"irm https:",[69,1180,1181],{"class":87},"\u002F\u002F",[69,1183,1184],{"class":479},"claude.ai",[69,1186,1187],{"class":87},"\u002F",[69,1189,1190],{"class":479},"install.ps1 ",[69,1192,1193],{"class":87},"|",[69,1195,836],{"class":479},[11,1197,1198],{},"Open a new terminal and check:",[59,1200,1202],{"className":61,"code":1201,"language":63,"meta":64,"style":64},"claude --version\n",[66,1203,1204],{"__ignoreMap":64},[69,1205,1206,1208],{"class":71,"line":72},[69,1207,658],{"class":75},[69,1209,1210],{"class":83}," --version\n",[21,1212,1214],{"id":1213},"option-2-npm","Option 2: npm",[11,1216,1217],{},"If Node.js 18 or later is installed:",[59,1219,1220],{"className":61,"code":842,"language":63,"meta":64,"style":64},[66,1221,1222],{"__ignoreMap":64},[69,1223,1224,1226,1228,1230],{"class":71,"line":72},[69,1225,849],{"class":75},[69,1227,852],{"class":79},[69,1229,855],{"class":83},[69,1231,858],{"class":79},[11,1233,1234,1235,1238],{},"Do not use ",[66,1236,1237],{},"sudo",". If npm cannot install global packages without root, configure a user-level prefix or use the native installer instead.",[21,1240,1242],{"id":1241},"first-sign-in","First sign-in",[11,1244,1245],{},"Change into a project folder and start the agent:",[59,1247,1249],{"className":61,"code":1248,"language":63,"meta":64,"style":64},"cd ~\u002Fprojects\u002Fmy-repo\nclaude\n",[66,1250,1251,1258],{"__ignoreMap":64},[69,1252,1253,1255],{"class":71,"line":72},[69,1254,128],{"class":83},[69,1256,1257],{"class":79}," ~\u002Fprojects\u002Fmy-repo\n",[69,1259,1260],{"class":71,"line":125},[69,1261,137],{"class":75},[11,1263,1264],{},"On first launch a browser window opens for sign-in. You can use a Claude account (Pro, Max, Team, Enterprise) or an Anthropic Console account (API billing). After that the agent is ready.",[11,1266,1267],{},"For scripts, CI and environments without a browser, set a key instead:",[59,1269,1271],{"className":61,"code":1270,"language":63,"meta":64,"style":64},"export ANTHROPIC_API_KEY=sk-ant-...\n",[66,1272,1273],{"__ignoreMap":64},[69,1274,1275,1278,1281,1284],{"class":71,"line":72},[69,1276,1277],{"class":87},"export",[69,1279,1280],{"class":479}," ANTHROPIC_API_KEY",[69,1282,1283],{"class":87},"=",[69,1285,1286],{"class":479},"sk-ant-...\n",[21,1288,1290],{"id":1289},"setting-up-the-project","Setting up the project",[11,1292,1293,1294,1296],{},"The most important step after installation is a ",[66,1295,143],{}," in the repository. It is loaded every session and should say how the project is built and tested, which conventions apply, and what the agent must not touch.",[59,1298,1303],{"className":1299,"code":1301,"language":1302},[1300],"language-text","> \u002Finit\n","text",[66,1304,1301],{"__ignoreMap":64},[11,1306,1307],{},"drafts one from what the agent finds in the repo. Edit it and commit it; it then applies to the whole team.",[11,1309,1310],{},"Other useful commands inside a session:",[158,1312,1313,1323],{},[161,1314,1315],{},[164,1316,1317,1320],{},[167,1318,1319],{},"Command",[167,1321,1322],{},"Purpose",[174,1324,1325,1335,1345,1354,1364],{},[164,1326,1327,1332],{},[179,1328,1329],{},[66,1330,1331],{},"\u002Fhelp",[179,1333,1334],{},"All commands",[164,1336,1337,1342],{},[179,1338,1339],{},[66,1340,1341],{},"\u002Fstatus",[179,1343,1344],{},"Version, account, model, context",[164,1346,1347,1351],{},[179,1348,1349],{},[66,1350,938],{},[179,1352,1353],{},"Show and authenticate MCP servers",[164,1355,1356,1361],{},[179,1357,1358],{},[66,1359,1360],{},"\u002Fpermissions",[179,1362,1363],{},"Allowed and blocked actions",[164,1365,1366,1371],{},[179,1367,1368],{},[66,1369,1370],{},"\u002Fclear",[179,1372,1373],{},"Clear the context",[21,1375,1377],{"id":1376},"editor-integration","Editor integration",[11,1379,1380],{},"Claude Code is editor-independent but integrates with the common ones:",[26,1382,1383,1389,1395],{},[29,1384,1385,1388],{},[32,1386,1387],{},"VS Code",": install the extension from the marketplace; diffs appear in the editor.",[29,1390,1391,1394],{},[32,1392,1393],{},"JetBrains"," (IntelliJ, PyCharm, WebStorm and the rest): plugin from the marketplace.",[29,1396,1397,1400],{},[32,1398,1399],{},"Any editor's terminal",": works everywhere, no plugin.",[21,1402,1404],{"id":1403},"common-errors","Common errors",[11,1406,1407,1412,1413,1416],{},[32,1408,1409,149],{},[66,1410,1411],{},"claude: command not found"," Restart the terminal so the PATH is refreshed. With npm, check that the global bin directory is on the PATH (",[66,1414,1415],{},"npm bin -g",").",[11,1418,1419,1422],{},[32,1420,1421],{},"Windows: shell commands fail."," Git for Windows is missing or not on the PATH. Restart the terminal after installing Git.",[11,1424,1425,1428,1429,1432],{},[32,1426,1427],{},"Sign-in fails on a corporate network."," Set proxy variables (",[66,1430,1431],{},"HTTPS_PROXY","); with TLS inspection, register the corporate certificate with Node or the system store.",[11,1434,1435,1438,1439,149],{},[32,1436,1437],{},"\"Rate limit\" or allowance exhausted."," Pro and Max have usage windows; API accounts have account limits. See ",[15,1440,1441],{"href":430},"Claude Code pricing",[21,1443,1445],{"id":1444},"running-through-an-eu-region","Running through an EU region",[11,1447,1448],{},"By default, requests go to Anthropic in the US. If your data protection rules require EU processing, run Claude Code through one of the supported cloud providers. The switch is environment variables, for example for Bedrock:",[59,1450,1452],{"className":61,"code":1451,"language":63,"meta":64,"style":64},"export CLAUDE_CODE_USE_BEDROCK=1\nexport AWS_REGION=eu-central-1\n",[66,1453,1454,1466],{"__ignoreMap":64},[69,1455,1456,1458,1461,1463],{"class":71,"line":72},[69,1457,1277],{"class":87},[69,1459,1460],{"class":479}," CLAUDE_CODE_USE_BEDROCK",[69,1462,1283],{"class":87},[69,1464,1465],{"class":83},"1\n",[69,1467,1468,1470,1473,1475],{"class":71,"line":125},[69,1469,1277],{"class":87},[69,1471,1472],{"class":479}," AWS_REGION",[69,1474,1283],{"class":87},[69,1476,1477],{"class":479},"eu-central-1\n",[11,1479,1480],{},"Equivalent variables exist for Vertex AI and Microsoft Foundry. Developers keep using the same tool; only the backend changes.",[21,1482,1484],{"id":1483},"running-through-a-gateway","Running through a gateway",[11,1486,1487],{},"For more than a handful of developers, the region alone is not enough. You need one place for model approvals, budgets, allowed MCP servers and the log. Claude Code supports a configurable base URL so that every request goes through a gateway:",[59,1489,1491],{"className":61,"code":1490,"language":63,"meta":64,"style":64},"export ANTHROPIC_BASE_URL=https:\u002F\u002Fhub.your-company.eu\nexport ANTHROPIC_AUTH_TOKEN=...\n",[66,1492,1493,1505],{"__ignoreMap":64},[69,1494,1495,1497,1500,1502],{"class":71,"line":72},[69,1496,1277],{"class":87},[69,1498,1499],{"class":479}," ANTHROPIC_BASE_URL",[69,1501,1283],{"class":87},[69,1503,1504],{"class":479},"https:\u002F\u002Fhub.your-company.eu\n",[69,1506,1507,1509,1512,1514],{"class":71,"line":125},[69,1508,1277],{"class":87},[69,1510,1511],{"class":479}," ANTHROPIC_AUTH_TOKEN",[69,1513,1283],{"class":87},[69,1515,1516],{"class":479},"...\n",[11,1518,1519,1520,149],{},"Walma AI Hub is such a gateway, running in your own Azure tenant in an EU region. Developers install a signed client for Windows and macOS that sets this configuration, and work with one key for Claude Code, Codex, Cursor and every approved MCP server. Policy is enforced in the gateway, not on the laptop. ",[15,1521,1522],{"href":670},"More about AI Hub",[679,1524,1525],{},"html pre.shiki code .sAwPA, html code.shiki .sAwPA{--shiki-default:#6A737D}html pre.shiki code .svObZ, html code.shiki .svObZ{--shiki-default:#B392F0}html pre.shiki code .sDLfK, html code.shiki .sDLfK{--shiki-default:#79B8FF}html pre.shiki code .sU2Wk, html code.shiki .sU2Wk{--shiki-default:#9ECBFF}html pre.shiki code .snl16, html code.shiki .snl16{--shiki-default:#F97583}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html pre.shiki code .s95oV, html code.shiki .s95oV{--shiki-default:#E1E4E8}",{"title":64,"searchDepth":125,"depth":134,"links":1527},[1528,1529,1530,1531,1532,1533,1534,1535,1536],{"id":1093,"depth":125,"text":1094},{"id":1136,"depth":125,"text":1137},{"id":1213,"depth":125,"text":1214},{"id":1241,"depth":125,"text":1242},{"id":1289,"depth":125,"text":1290},{"id":1376,"depth":125,"text":1377},{"id":1403,"depth":125,"text":1404},{"id":1444,"depth":125,"text":1445},{"id":1483,"depth":125,"text":1484},"Step-by-step installation of Claude Code with the native installer or npm, first sign-in, project setup with CLAUDE.md, the VS Code and JetBrains integrations, common errors, and how to run it through an EU region or a gateway.",[1539,1542,1545,1548,1551],{"q":1540,"a":1541},"Do I need Node.js to install Claude Code?","Not any more. The native installer is self-contained. Only the npm route requires Node.js 18 or later.",{"q":1543,"a":1544},"How do I update Claude Code?","The native installer updates itself in the background. With npm, run npm install -g @anthropic-ai\u002Fclaude-code again. Inside a session, \u002Fstatus shows the current version.",{"q":1546,"a":1547},"Claude Code won't start on Windows. What should I check?","Make sure Git for Windows is installed; Claude Code uses its Bash for shell commands. Restart the terminal after installing so the PATH is refreshed. As an alternative, Claude Code runs reliably inside WSL.",{"q":1549,"a":1550},"Can I use Claude Code without an Anthropic account?","Yes, with an API key via the ANTHROPIC_API_KEY environment variable, or through Amazon Bedrock, Google Vertex AI or Microsoft Foundry using those clouds' credentials.",{"q":1552,"a":1553},"How do I use Claude Code in VS Code?","Install the Claude Code extension from the VS Code marketplace; diffs then appear in the editor and you can start sessions from the sidebar. Running the CLI in VS Code's integrated terminal works too, with no extension.",{},"\u002Fguides\u002Fclaude-code\u002Fhow-to-install-claude-code",{"title":1080,"description":1537},"guides\u002Fclaude-code\u002Fhow-to-install-claude-code","claude-code-install","YGHP6DTDshBdF9TM0FMpxWREc6MWSDQvZG6vgvmFWzU",{"id":1561,"title":1562,"author":6,"body":1563,"date":1052,"description":1785,"extension":695,"faq":1786,"meta":1801,"navigation":713,"order":125,"path":1802,"readTime":1803,"seo":1804,"stem":1805,"topic":718,"translationId":1806,"updated":1052,"__hash__":1807},"guides\u002Fguides\u002Fclaude-code\u002Fclaude-code-pricing.md","Is Claude Code free? Claude Code pricing and cost control explained",{"type":8,"value":1564,"toc":1778},[1565,1570,1573,1577,1676,1680,1683,1686,1690,1693,1696,1716,1719,1723,1729,1734,1737,1741,1744,1771],[11,1566,1567,1568,149],{},"Claude Code has no price of its own. It is a tool that runs on one of Anthropic's subscriptions or on pay-per-token API billing. What you pay therefore depends on which route you choose, which model you use, and how hard the agent works. This guide explains the options, puts the numbers in context, and shows how to keep costs under control across a team. For what Claude Code is, start with the ",[15,1569,18],{"href":17},[11,1571,1572],{},"A note up front: prices change. The figures here are as of September 2026 in US dollars; Anthropic's pricing page has the current amounts and local currency.",[21,1574,1576],{"id":1575},"the-options-at-a-glance","The options at a glance",[158,1578,1579,1595],{},[161,1580,1581],{},[164,1582,1583,1586,1589,1592],{},[167,1584,1585],{},"Option",[167,1587,1588],{},"Who it is for",[167,1590,1591],{},"Billing",[167,1593,1594],{},"Claude Code included",[174,1596,1597,1610,1624,1638,1651,1663],{},[164,1598,1599,1601,1604,1607],{},[179,1600,881],{},[179,1602,1603],{},"Individuals, occasional use",[179,1605,1606],{},"~20 USD\u002Fmonth",[179,1608,1609],{},"Yes, with a usage window",[164,1611,1612,1615,1618,1621],{},[179,1613,1614],{},"Max 5x",[179,1616,1617],{},"Developers using it daily",[179,1619,1620],{},"~100 USD\u002Fmonth",[179,1622,1623],{},"Yes, ~5× the Pro allowance",[164,1625,1626,1629,1632,1635],{},[179,1627,1628],{},"Max 20x",[179,1630,1631],{},"Heavy users, parallel sessions",[179,1633,1634],{},"~200 USD\u002Fmonth",[179,1636,1637],{},"Yes, ~20× the Pro allowance",[164,1639,1640,1642,1645,1648],{},[179,1641,407],{},[179,1643,1644],{},"Small and mid-size teams, central admin",[179,1646,1647],{},"Per seat per month",[179,1649,1650],{},"In the premium seats",[164,1652,1653,1655,1658,1660],{},[179,1654,418],{},[179,1656,1657],{},"Large organisations, SSO, audit, contracts",[179,1659,421],{},[179,1661,1662],{},"Yes",[164,1664,1665,1667,1670,1673],{},[179,1666,899],{},[179,1668,1669],{},"CI, custom agents, full control",[179,1671,1672],{},"Per token",[179,1674,1675],{},"Yes, via API key",[21,1677,1679],{"id":1678},"what-usage-allowance-means","What \"usage allowance\" means",[11,1681,1682],{},"Pro and Max do not bill per token. They give you an allowance within rolling time windows. When it is used up, the agent pauses until the next window opens. There are additional weekly caps for very heavy users.",[11,1684,1685],{},"How far the allowance stretches depends on the model and the kind of work. Short, targeted tasks in a small repository use little. Long sessions where the agent reads large files, reruns tests and calls many tools use a lot. Rule of thumb: Pro covers a few hours a day with breaks, Max 5x a full working day, Max 20x parallel sessions or very large codebases.",[21,1687,1689],{"id":1688},"api-billing-what-a-developer-spends","API billing: what a developer spends",[11,1691,1692],{},"Through the API you pay per input and output token, with different prices per model. Sonnet-class models are considerably cheaper than Opus-class ones, and prompt caching cuts the cost of repeated inputs substantially; Claude Code uses it automatically.",[11,1694,1695],{},"From the rollouts we support, the orders of magnitude are:",[26,1697,1698,1704,1710],{},[29,1699,1700,1703],{},[32,1701,1702],{},"Occasional use"," (a few tasks a week): single to low double digits in dollars per month.",[29,1705,1706,1709],{},[32,1707,1708],{},"Daily use"," (several hours): mid double digits to low triple digits.",[29,1711,1712,1715],{},[32,1713,1714],{},"Heavy use with Opus models and large contexts",": several hundred dollars is possible.",[11,1717,1718],{},"The point that surprises finance teams: the spread between developers is wide, and a single session can cost many times the daily average. Without budget limits you find out on the monthly invoice.",[21,1720,1722],{"id":1721},"subscription-or-api","Subscription or API?",[11,1724,1725,1728],{},[32,1726,1727],{},"Subscription"," when developers use the tool interactively and you want predictable per-head cost. Team and Enterprise add central administration, SSO and the commercial terms under which data is not used for training.",[11,1730,1731,1733],{},[32,1732,899],{}," when Claude Code runs in CI pipelines, automations or your own agents, when you want to choose the backend yourself (an EU region via Bedrock, Vertex AI or Microsoft Foundry), or when you operate a gateway with your own budgets.",[11,1735,1736],{},"Many companies combine the two: subscriptions for interactive use, API for automation.",[21,1738,1740],{"id":1739},"controlling-cost-across-a-team","Controlling cost across a team",[11,1742,1743],{},"Four measures that make the difference in practice:",[1745,1746,1747,1753,1759,1765],"ol",{},[29,1748,1749,1752],{},[32,1750,1751],{},"Budgets and hard caps"," per user, team and project. Not a report after the fact, but a limit the agent cannot exceed.",[29,1754,1755,1758],{},[32,1756,1757],{},"Model policies."," Not every task needs the most expensive model. A policy that defaults to a cheaper model and unlocks the expensive one on request often halves cost with no noticeable quality loss.",[29,1760,1761,1764],{},[32,1762,1763],{},"A forecast to month end"," per model and team, based on actual usage over the last weeks rather than a straight line.",[29,1766,1767,1770],{},[32,1768,1769],{},"A log per call",": who, when, which model, how many tokens, which repository. Without it every cost discussion is a guess.",[11,1772,1773,1774,1777],{},"These four are why companies put a gateway in front of Claude Code. Walma AI Hub enforces budgets and model policy in the gateway, forecasts cost per model to month end, and logs every call, for Claude Code, Codex, Cursor and the approved MCP servers, running in the customer's own Azure tenant in an EU region. There is a cost calculator on the home page; for a concrete scenario, a ",[15,1775,1776],{"href":670},"walkthrough with an engineer"," is the fastest route.",{"title":64,"searchDepth":125,"depth":134,"links":1779},[1780,1781,1782,1783,1784],{"id":1575,"depth":125,"text":1576},{"id":1678,"depth":125,"text":1679},{"id":1688,"depth":125,"text":1689},{"id":1721,"depth":125,"text":1722},{"id":1739,"depth":125,"text":1740},"What Claude Code costs: the differences between Pro, Max, Team, Enterprise and API billing, what a usage allowance means in practice, how much a developer typically spends per month, and how to keep costs under control across a team.",[1787,1789,1792,1795,1798],{"q":1059,"a":1788},"No. There is no permanently free tier for Claude Code. It is included in the Pro, Max, Team and Enterprise subscriptions, or billed per token through the API. New API accounts sometimes receive a small trial credit.",{"q":1790,"a":1791},"How much does Claude Code cost per month?","Through a subscription: Pro is around 20 US dollars per month, Max around 100 or 200 depending on tier, and Team and Enterprise are priced per seat. Through the API you pay per token, which for a developer using the agent daily typically lands in the tens to low hundreds of dollars per month. Check Anthropic's pricing page for current amounts.",{"q":1793,"a":1794},"Is Claude Pro enough for Claude Code?","For occasional use, yes. Developers who run the agent for several hours a day regularly hit Pro's usage window and are better served by Max.",{"q":1796,"a":1797},"Is Claude Code cheaper through the API or a subscription?","For light use the API is cheaper; for heavy daily use a Max subscription usually is. Many companies combine subscriptions for interactive work with API billing for CI and automation.",{"q":1799,"a":1800},"How do we control Claude Code costs across a team?","Budgets and hard caps per user and team, model policies that default to cheaper models, a forecast to month end, and a log per call. All of that is easiest to enforce in a gateway that every request passes through.",{},"\u002Fguides\u002Fclaude-code\u002Fclaude-code-pricing","9 min read",{"title":1562,"description":1785},"guides\u002Fclaude-code\u002Fclaude-code-pricing","claude-code-pricing","UbbSncrZisIzzlXrbndOFfAg1MHRwUd6GVh-NYdrtUA",{"id":4,"title":5,"author":6,"body":1809,"date":693,"description":694,"extension":695,"faq":2262,"meta":2268,"navigation":713,"order":134,"path":714,"readTime":715,"seo":2269,"stem":717,"topic":718,"translationId":719,"updated":693,"__hash__":720},{"type":8,"value":1810,"toc":2251},[1811,1815,1817,1827,1829,1831,1837,1853,1857,1861,1887,1893,1895,1897,1945,1947,1949,1951,1953,1957,1977,1981,2005,2027,2029,2031,2033,2087,2091,2093,2095,2111,2147,2161,2175,2177,2183,2193,2195,2197,2233,2235,2243,2249],[11,1812,13,1813,19],{},[15,1814,18],{"href":17},[21,1816,24],{"id":23},[26,1818,1819,1823],{},[29,1820,1821,35],{},[32,1822,34],{},[29,1824,1825,41],{},[32,1826,40],{},[11,1828,44],{},[21,1830,48],{"id":47},[11,1832,1833,53,1835,57],{},[32,1834,34],{},[32,1836,56],{},[59,1838,1839],{"className":61,"code":62,"language":63,"meta":64,"style":64},[66,1840,1841],{"__ignoreMap":64},[69,1842,1843,1845,1847,1849,1851],{"class":71,"line":72},[69,1844,76],{"class":75},[69,1846,80],{"class":79},[69,1848,84],{"class":83},[69,1850,88],{"class":87},[69,1852,91],{"class":75},[11,1854,94,1855,98],{},[66,1856,97],{},[11,1858,1859,103],{},[32,1860,40],{},[59,1862,1863],{"className":61,"code":106,"language":63,"meta":64,"style":64},[66,1864,1865,1877,1883],{"__ignoreMap":64},[69,1866,1867,1869,1871,1873,1875],{"class":71,"line":72},[69,1868,76],{"class":75},[69,1870,115],{"class":83},[69,1872,118],{"class":79},[69,1874,88],{"class":87},[69,1876,91],{"class":75},[69,1878,1879,1881],{"class":71,"line":125},[69,1880,128],{"class":83},[69,1882,131],{"class":79},[69,1884,1885],{"class":71,"line":134},[69,1886,137],{"class":75},[11,1888,140,1889,144,1891,149],{},[66,1890,143],{},[15,1892,148],{"href":147},[11,1894,152],{},[21,1896,156],{"id":155},[158,1898,1899,1909],{},[161,1900,1901],{},[164,1902,1903,1905,1907],{},[167,1904],{},[167,1906,34],{},[167,1908,40],{},[174,1910,1911,1919,1927,1937],{},[164,1912,1913,1915,1917],{},[179,1914,181],{},[179,1916,184],{},[179,1918,187],{},[164,1920,1921,1923,1925],{},[179,1922,192],{},[179,1924,195],{},[179,1926,198],{},[164,1928,1929,1931,1933],{},[179,1930,203],{},[179,1932,206],{},[179,1934,209,1935,213],{},[66,1936,212],{},[164,1938,1939,1941,1943],{},[179,1940,218],{},[179,1942,221],{},[179,1944,224],{},[11,1946,227],{},[11,1948,230],{},[21,1950,234],{"id":233},[11,1952,237],{},[11,1954,1955],{},[32,1956,34],{},[26,1958,1959,1963,1967],{},[29,1960,1961,249],{},[32,1962,248],{},[29,1964,1965,255],{},[32,1966,254],{},[29,1968,1969,261,1971,265,1973,269,1975,273],{},[32,1970,260],{},[32,1972,264],{},[66,1974,268],{},[66,1976,272],{},[11,1978,1979],{},[32,1980,40],{},[26,1982,1983,1987,1993,1999],{},[29,1984,1985,285],{},[32,1986,284],{},[29,1988,1989,265,1991,294],{},[32,1990,290],{},[66,1992,293],{},[29,1994,1995,300,1997,304],{},[32,1996,299],{},[66,1998,303],{},[29,2000,2001,309,2003,313],{},[32,2002,260],{},[32,2004,312],{},[59,2006,2007],{"className":61,"code":316,"language":63,"meta":64,"style":64},[66,2008,2009],{"__ignoreMap":64},[69,2010,2011,2013,2015,2017,2019,2021,2023,2025],{"class":71,"line":72},[69,2012,323],{"class":75},[69,2014,326],{"class":79},[69,2016,329],{"class":79},[69,2018,332],{"class":83},[69,2020,88],{"class":87},[69,2022,337],{"class":75},[69,2024,340],{"class":83},[69,2026,343],{"class":79},[11,2028,346],{},[21,2030,350],{"id":349},[11,2032,353],{},[158,2034,2035,2045],{},[161,2036,2037],{},[164,2038,2039,2041,2043],{},[167,2040,362],{},[167,2042,34],{},[167,2044,367],{},[174,2046,2047,2055,2063,2071,2079],{},[164,2048,2049,2051,2053],{},[179,2050,374],{},[179,2052,377],{},[179,2054,380],{},[164,2056,2057,2059,2061],{},[179,2058,385],{},[179,2060,388],{},[179,2062,391],{},[164,2064,2065,2067,2069],{},[179,2066,396],{},[179,2068,399],{},[179,2070,402],{},[164,2072,2073,2075,2077],{},[179,2074,407],{},[179,2076,410],{},[179,2078,413],{},[164,2080,2081,2083,2085],{},[179,2082,418],{},[179,2084,421],{},[179,2086,424],{},[11,2088,427,2089,432],{},[15,2090,431],{"href":430},[21,2092,436],{"id":435},[11,2094,439],{},[11,2096,2097,445,2099,449,2101,453,2103,457,2105,461,2107,261,2109,149],{},[32,2098,444],{},[66,2100,448],{},[66,2102,452],{},[66,2104,456],{},[66,2106,460],{},[15,2108,34],{"href":464},[15,2110,40],{"href":467},[59,2112,2113],{"className":470,"code":471,"language":472,"meta":64,"style":64},[66,2114,2115,2119,2125,2139,2143],{"__ignoreMap":64},[69,2116,2117],{"class":71,"line":72},[69,2118,480],{"class":479},[69,2120,2121,2123],{"class":71,"line":125},[69,2122,485],{"class":83},[69,2124,488],{"class":479},[69,2126,2127,2129,2131,2133,2135,2137],{"class":71,"line":134},[69,2128,493],{"class":83},[69,2130,496],{"class":479},[69,2132,499],{"class":83},[69,2134,502],{"class":479},[69,2136,505],{"class":79},[69,2138,508],{"class":479},[69,2140,2141],{"class":71,"line":511},[69,2142,514],{"class":479},[69,2144,2145],{"class":71,"line":517},[69,2146,520],{"class":479},[11,2148,2149,526,2151,530,2153,261,2155,537,2157,261,2159,149],{},[32,2150,525],{},[66,2152,529],{},[66,2154,533],{},[66,2156,536],{},[15,2158,541],{"href":540},[15,2160,545],{"href":544},[11,2162,2163,551,2165,554,2167,557,2169,261,2171,562,2173,149],{},[32,2164,550],{},[66,2166,143],{},[66,2168,272],{},[66,2170,268],{},[66,2172,272],{},[66,2174,272],{},[21,2176,568],{"id":567},[11,2178,2179,574,2181,578],{},[32,2180,573],{},[32,2182,577],{},[11,2184,2185,583,2187,587,2189,591,2191,595],{},[32,2186,40],{},[66,2188,586],{},[66,2190,590],{},[66,2192,594],{},[11,2194,598],{},[21,2196,602],{"id":601},[158,2198,2199,2207],{},[161,2200,2201],{},[164,2202,2203,2205],{},[167,2204,611],{},[167,2206,614],{},[174,2208,2209,2215,2221,2227],{},[164,2210,2211,2213],{},[179,2212,621],{},[179,2214,624],{},[164,2216,2217,2219],{},[179,2218,629],{},[179,2220,632],{},[164,2222,2223,2225],{},[179,2224,637],{},[179,2226,640],{},[164,2228,2229,2231],{},[179,2230,645],{},[179,2232,648],{},[21,2234,652],{"id":651},[11,2236,655,2237,659,2239,269,2241,664],{},[66,2238,658],{},[66,2240,272],{},[66,2242,533],{},[11,2244,667,2245,672,2247,677],{},[15,2246,671],{"href":670},[15,2248,676],{"href":675},[679,2250,681],{},{"title":64,"searchDepth":125,"depth":134,"links":2252},[2253,2254,2255,2256,2257,2258,2259,2260,2261],{"id":23,"depth":125,"text":24},{"id":47,"depth":125,"text":48},{"id":155,"depth":125,"text":156},{"id":233,"depth":125,"text":234},{"id":349,"depth":125,"text":350},{"id":435,"depth":125,"text":436},{"id":567,"depth":125,"text":568},{"id":601,"depth":125,"text":602},{"id":651,"depth":125,"text":652},[2263,2264,2265,2266,2267],{"q":698,"a":699},{"q":701,"a":702},{"q":704,"a":705},{"q":707,"a":708},{"q":710,"a":711},{},{"title":5,"description":694},{"id":2271,"title":2272,"author":6,"body":2273,"date":693,"description":3126,"extension":695,"faq":3127,"meta":3143,"navigation":713,"order":511,"path":3144,"readTime":715,"seo":3145,"stem":3146,"topic":718,"translationId":719,"updated":693,"__hash__":3147},"guides\u002Fguides\u002Fclaude-code\u002Fclaude-code-vs-codex.md","Claude Code vs Codex: an honest comparison for teams (2026)",{"type":8,"value":2274,"toc":3115},[2275,2278,2284,2288,2396,2400,2403,2459,2469,2474,2476,2479,2504,2523,2526,2530,2533,2607,2612,2616,2619,2650,2672,2706,2713,2784,2787,2791,2794,2799,2847,2871,2910,2938,2957,2961,2964,2969,2978,3062,3065,3069,3072,3086,3090,3113],[11,2276,2277],{},"Claude Code and OpenAI Codex have converged. Both started as terminal agents, both now run in the IDE, in a desktop app and in the cloud, and both read project instructions, call MCP servers and load skills. So \"which one is better\" is the wrong question for most teams. The useful questions are which models you want, how much autonomy each tool grants by default, and how either one fits your cloud, your budget and your compliance rules.",[11,2279,2280,2281,149],{},"This guide compares the two as of October 2026, based on the vendors' own documentation. If you are new to Anthropic's tool, start with the ",[15,2282,2283],{"href":17},"Claude Code overview",[21,2285,2287],{"id":2286},"at-a-glance","At a glance",[158,2289,2290,2301],{},[161,2291,2292],{},[164,2293,2294,2296,2298],{},[167,2295],{},[167,2297,40],{},[167,2299,2300],{},"Codex",[174,2302,2303,2314,2325,2335,2352,2367,2385],{},[164,2304,2305,2308,2311],{},[179,2306,2307],{},"Vendor",[179,2309,2310],{},"Anthropic",[179,2312,2313],{},"OpenAI",[164,2315,2316,2319,2322],{},[179,2317,2318],{},"Surfaces",[179,2320,2321],{},"Terminal, VS Code, JetBrains, desktop app, web, Slack, GitHub Actions",[179,2323,2324],{},"CLI, IDE extension, desktop app, web\u002Fcloud, GitHub",[164,2326,2327,2329,2332],{},[179,2328,156],{},[179,2330,2331],{},"Opus, Sonnet, Haiku (Opus 5.5 is the default on paid plans)",[179,2333,2334],{},"GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna",[164,2336,2337,2340,2348],{},[179,2338,2339],{},"Project instructions",[179,2341,2342,2344,2345,2347],{},[66,2343,143],{}," (reads ",[66,2346,272],{}," too)",[179,2349,2350],{},[66,2351,272],{},[164,2353,2354,2357,2362],{},[179,2355,2356],{},"Skills folder",[179,2358,2359],{},[66,2360,2361],{},".claude\u002Fskills",[179,2363,2364],{},[66,2365,2366],{},".agents\u002Fskills",[164,2368,2369,2372,2378],{},[179,2370,2371],{},"Local sandbox",[179,2373,2374,2375],{},"Opt-in, ",[66,2376,2377],{},"\u002Fsandbox",[179,2379,2380,2381,2384],{},"On by default (",[66,2382,2383],{},"workspace-write",")",[164,2386,2387,2390,2393],{},[179,2388,2389],{},"Cheapest paid plan",[179,2391,2392],{},"Pro, 20 USD\u002Fmonth",[179,2394,2395],{},"Plus, 20 USD\u002Fmonth (also on Free and Go)",[21,2397,2399],{"id":2398},"workflow-and-surfaces","Workflow and surfaces",[11,2401,2402],{},"Day to day, the two feel alike. You start the agent in a repository, describe a task, and it searches the code, plans, edits files and runs tests.",[59,2404,2406],{"className":61,"code":2405,"language":63,"meta":64,"style":64},"# Claude Code\ncurl -fsSL https:\u002F\u002Fclaude.ai\u002Finstall.sh | bash\nclaude\n\n# Codex\ncurl -fsSL https:\u002F\u002Fchatgpt.com\u002Fcodex\u002Finstall.sh | sh\ncodex\n",[66,2407,2408,2413,2425,2429,2433,2438,2453],{"__ignoreMap":64},[69,2409,2410],{"class":71,"line":72},[69,2411,2412],{"class":800},"# Claude Code\n",[69,2414,2415,2417,2419,2421,2423],{"class":71,"line":125},[69,2416,76],{"class":75},[69,2418,115],{"class":83},[69,2420,118],{"class":79},[69,2422,88],{"class":87},[69,2424,91],{"class":75},[69,2426,2427],{"class":71,"line":134},[69,2428,137],{"class":75},[69,2430,2431],{"class":71,"line":511},[69,2432,818],{"emptyLinePlaceholder":713},[69,2434,2435],{"class":71,"line":517},[69,2436,2437],{"class":800},"# Codex\n",[69,2439,2441,2443,2445,2448,2450],{"class":71,"line":2440},6,[69,2442,76],{"class":75},[69,2444,115],{"class":83},[69,2446,2447],{"class":79}," https:\u002F\u002Fchatgpt.com\u002Fcodex\u002Finstall.sh",[69,2449,88],{"class":87},[69,2451,2452],{"class":75}," sh\n",[69,2454,2456],{"class":71,"line":2455},7,[69,2457,2458],{"class":75},"codex\n",[11,2460,2461,2462,261,2465,2468],{},"Both have a non-interactive mode for scripts and CI: ",[66,2463,2464],{},"claude -p \"...\"",[66,2466,2467],{},"codex exec \"...\"",". Both run long tasks in the cloud while your laptop is closed, Claude Code on the web and in the Claude mobile app, Codex in its web and cloud surface. Claude Code adds scheduled routines, a Slack integration that turns a bug report into a pull request, and subagents that split a task in parallel. Codex is tightly tied to the ChatGPT account and apps, which matters if your organisation already lives in ChatGPT.",[11,2470,2471,2472,149],{},"Installation details for Claude Code, including Windows, are in ",[15,2473,867],{"href":147},[21,2475,156],{"id":155},[11,2477,2478],{},"This is the real difference, and the reason the comparison never settles.",[11,2480,2481,2483,2484,2487,2488,2491,2492,2495,2496,2499,2500,2503],{},[32,2482,40],{}," runs Anthropic's models. The ",[66,2485,2486],{},"opus"," alias currently maps to Opus 5.5 on Anthropic's API, ",[66,2489,2490],{},"sonnet"," to Sonnet 5.5, and ",[66,2493,2494],{},"haiku"," covers quick, cheap work. Switch with ",[66,2497,2498],{},"\u002Fmodel"," in a session or ",[66,2501,2502],{},"claude --model sonnet"," at startup. Note that the mapping differs by provider: on Amazon Bedrock and Microsoft Foundry the same aliases point to older versions, so check what your cloud actually serves.",[11,2505,2506,2508,2509,2511,2512,2515,2516,2519,2520,149],{},[32,2507,2300],{}," runs OpenAI's models. OpenAI's docs recommend GPT-6 Astra as the most capable, GPT-6.1 Sol as near-Astra at lower cost, and GPT-6 Luna for focused, efficient tasks. GPT-5.5 retires on 14 October 2026. Switch with ",[66,2510,2498],{}," or ",[66,2513,2514],{},"codex -m gpt-6.1-sol",", or set ",[66,2517,2518],{},"model"," in ",[66,2521,2522],{},"~\u002F.codex\u002Fconfig.toml",[11,2524,2525],{},"In practice developers develop strong preferences per task type: one model plans a refactor better, the other writes the tests faster, and the ranking shifts with every release. We are not going to publish a winner here, because any benchmark in a guide like this is out of date within a quarter. Run both on a week of your own tickets and look at the diffs.",[21,2527,2529],{"id":2528},"pricing-and-plans","Pricing and plans",[11,2531,2532],{},"Neither tool is sold on its own. Both come with a chat subscription or with per-token API billing.",[158,2534,2535,2546],{},[161,2536,2537],{},[164,2538,2539,2542,2544],{},[167,2540,2541],{},"Plan",[167,2543,40],{},[167,2545,2300],{},[174,2547,2548,2558,2568,2578,2588,2598],{},[164,2549,2550,2552,2555],{},[179,2551,374],{},[179,2553,2554],{},"Not included",[179,2556,2557],{},"Included (ChatGPT Free)",[164,2559,2560,2562,2565],{},[179,2561,385],{},[179,2563,2564],{},"Pro, 20 USD\u002Fmonth (17 billed annually)",[179,2566,2567],{},"Go, 8 USD\u002Fmonth; Plus, 20 USD\u002Fmonth",[164,2569,2570,2572,2575],{},[179,2571,396],{},[179,2573,2574],{},"Max, from 100 USD\u002Fmonth",[179,2576,2577],{},"Pro, from 100 USD\u002Fmonth",[164,2579,2580,2582,2585],{},[179,2581,407],{},[179,2583,2584],{},"Team standard seat 20 USD (annual) or 25 USD (monthly); premium seat 100\u002F125 USD",[179,2586,2587],{},"Business, 20 USD\u002Fuser (annual) or 25 USD (monthly)",[164,2589,2590,2592,2595],{},[179,2591,418],{},[179,2593,2594],{},"20 USD\u002Fseat\u002Fmonth plus usage at API rates",[179,2596,2597],{},"Custom pricing",[164,2599,2600,2603,2605],{},[179,2601,2602],{},"API key",[179,2604,1672],{},[179,2606,1672],{},[11,2608,2609,2610,149],{},"Usage limits are where the real cost hides. Both vendors meter subscription usage in rolling windows, and both say consumption depends heavily on model and task size. A developer who runs the agent for hours a day will outgrow the 20 dollar tier on either side. For Claude Code numbers and cost controls, see ",[15,2611,1441],{"href":430},[21,2613,2615],{"id":2614},"sandboxing-and-approvals","Sandboxing and approvals",[11,2617,2618],{},"Here the defaults differ, and it is worth understanding before you roll either out.",[11,2620,2621,2624,2625,269,2628,2630,2631,2634,2635,2637,2638,2641,2642,2645,2646,2649],{},[32,2622,2623],{},"Codex sandboxes by default."," It has three sandbox modes: ",[66,2626,2627],{},"read-only",[66,2629,2383],{}," (the default for local work: edit the workspace, run local commands inside it) and ",[66,2632,2633],{},"danger-full-access",". In ",[66,2636,2383],{},", outbound network access is off unless you opt in. Approval policies decide when the agent stops to ask: ",[66,2639,2640],{},"on-request"," asks when it needs to go beyond the sandbox, ",[66,2643,2644],{},"never"," does not stop. Enforcement is native: Seatbelt on macOS, ",[66,2647,2648],{},"bubblewrap"," on Linux and WSL2.",[59,2651,2653],{"className":61,"code":2652,"language":63,"meta":64,"style":64},"codex --sandbox workspace-write --ask-for-approval on-request\n",[66,2654,2655],{"__ignoreMap":64},[69,2656,2657,2660,2663,2666,2669],{"class":71,"line":72},[69,2658,2659],{"class":75},"codex",[69,2661,2662],{"class":83}," --sandbox",[69,2664,2665],{"class":79}," workspace-write",[69,2667,2668],{"class":83}," --ask-for-approval",[69,2670,2671],{"class":79}," on-request\n",[11,2673,2674,2677,2678,269,2681,2684,2685,2688,2689,261,2692,2695,2696,2698,2699,2701,2702,2705],{},[32,2675,2676],{},"Claude Code asks by default and sandboxes on request."," Its default mode prompts before edits and shell commands. Other modes are ",[66,2679,2680],{},"acceptEdits",[66,2682,2683],{},"plan"," (read-only exploration), ",[66,2686,2687],{},"auto"," (a classifier reviews actions instead of you), ",[66,2690,2691],{},"dontAsk",[66,2693,2694],{},"bypassPermissions",". The OS-level sandbox for shell commands uses the same Seatbelt and ",[66,2697,2648],{}," mechanisms but is off by default: run ",[66,2700,2377],{}," or set ",[66,2703,2704],{},"sandbox.enabled"," in settings. It covers shell commands only; file tools, MCP servers and hooks run outside it.",[11,2707,2708,2709,2712],{},"For a team, Claude Code's strength is central policy. Administrators deploy managed settings that users cannot override, including deny rules and ",[66,2710,2711],{},"disableBypassPermissionsMode",":",[59,2714,2716],{"className":470,"code":2715,"language":472,"meta":64,"style":64},"{\n  \"permissions\": {\n    \"deny\": [\"Read(.\u002F.env)\", \"Bash(curl:*)\"],\n    \"disableBypassPermissionsMode\": \"disable\"\n  },\n  \"sandbox\": { \"enabled\": true }\n}\n",[66,2717,2718,2722,2729,2748,2758,2763,2780],{"__ignoreMap":64},[69,2719,2720],{"class":71,"line":72},[69,2721,480],{"class":479},[69,2723,2724,2727],{"class":71,"line":125},[69,2725,2726],{"class":83},"  \"permissions\"",[69,2728,488],{"class":479},[69,2730,2731,2734,2737,2740,2742,2745],{"class":71,"line":134},[69,2732,2733],{"class":83},"    \"deny\"",[69,2735,2736],{"class":479},": [",[69,2738,2739],{"class":79},"\"Read(.\u002F.env)\"",[69,2741,269],{"class":479},[69,2743,2744],{"class":79},"\"Bash(curl:*)\"",[69,2746,2747],{"class":479},"],\n",[69,2749,2750,2753,2755],{"class":71,"line":511},[69,2751,2752],{"class":83},"    \"disableBypassPermissionsMode\"",[69,2754,502],{"class":479},[69,2756,2757],{"class":79},"\"disable\"\n",[69,2759,2760],{"class":71,"line":517},[69,2761,2762],{"class":479},"  },\n",[69,2764,2765,2768,2770,2773,2775,2778],{"class":71,"line":2440},[69,2766,2767],{"class":83},"  \"sandbox\"",[69,2769,496],{"class":479},[69,2771,2772],{"class":83},"\"enabled\"",[69,2774,502],{"class":479},[69,2776,2777],{"class":83},"true",[69,2779,508],{"class":479},[69,2781,2782],{"class":71,"line":2455},[69,2783,520],{"class":479},[11,2785,2786],{},"Our view: Codex has the safer out-of-the-box default for an individual developer; Claude Code gives an organisation finer, enforceable rules. Either way, set the policy centrally rather than trusting each laptop.",[21,2788,2790],{"id":2789},"mcp-skills-and-instruction-files","MCP, skills and instruction files",[11,2792,2793],{},"The ecosystems have largely standardised, which is good news if you run both.",[11,2795,2796,2798],{},[32,2797,444],{}," Both are MCP clients for stdio and streamable HTTP servers, with OAuth for remote ones. The same server, added to each:",[59,2800,2802],{"className":61,"code":2801,"language":63,"meta":64,"style":64},"claude mcp add context7 -- npx -y @upstash\u002Fcontext7-mcp\ncodex mcp add context7 -- npx -y @upstash\u002Fcontext7-mcp\n",[66,2803,2804,2829],{"__ignoreMap":64},[69,2805,2806,2808,2811,2814,2817,2820,2823,2826],{"class":71,"line":72},[69,2807,658],{"class":75},[69,2809,2810],{"class":79}," mcp",[69,2812,2813],{"class":79}," add",[69,2815,2816],{"class":79}," context7",[69,2818,2819],{"class":83}," --",[69,2821,2822],{"class":79}," npx",[69,2824,2825],{"class":83}," -y",[69,2827,2828],{"class":79}," @upstash\u002Fcontext7-mcp\n",[69,2830,2831,2833,2835,2837,2839,2841,2843,2845],{"class":71,"line":125},[69,2832,2659],{"class":75},[69,2834,2810],{"class":79},[69,2836,2813],{"class":79},[69,2838,2816],{"class":79},[69,2840,2819],{"class":83},[69,2842,2822],{"class":79},[69,2844,2825],{"class":83},[69,2846,2828],{"class":79},[11,2848,2849,2850,2852,2853,2856,2857,2859,2860,2863,2864,2867,2868,149],{},"Claude Code stores project servers in ",[66,2851,460],{},"; Codex uses ",[66,2854,2855],{},"[mcp_servers.\u003Cname>]"," blocks in ",[66,2858,2522],{}," or a project ",[66,2861,2862],{},".codex\u002Fconfig.toml",", and ",[66,2865,2866],{},"codex mcp login \u003Cname>"," for OAuth. More in ",[15,2869,2870],{"href":467},"Add an MCP server to Claude Code",[11,2872,2873,2875,2876,2878,2879,2882,2883,2886,2887,261,2889,2892,2893,261,2895,2898,2899,2511,2902,2905,2906,2909],{},[32,2874,525],{}," Both follow the open Agent Skills standard: a folder with a ",[66,2877,529],{}," holding a ",[66,2880,2881],{},"name",", a ",[66,2884,2885],{},"description"," and instructions. Claude Code reads ",[66,2888,2361],{},[66,2890,2891],{},"~\u002F.claude\u002Fskills","; Codex reads ",[66,2894,2366],{},[66,2896,2897],{},"~\u002F.agents\u002Fskills",". Claude Code adds its own frontmatter fields (such as ",[66,2900,2901],{},"context: fork",[66,2903,2904],{},"disable-model-invocation",") that Codex will ignore, so keep shared skills to the standard fields. See ",[15,2907,2908],{"href":544},"Claude skills vs MCP"," for when each fits.",[11,2911,2912,2915,2916,2918,2919,2922,2923,2925,2926,2928,2929,2931,2932,2934,2935,2937],{},[32,2913,2914],{},"AGENTS.md vs CLAUDE.md."," Codex reads ",[66,2917,272],{}," from ",[66,2920,2921],{},"~\u002F.codex",", then from the git root down to the current directory, capped at 32 KiB by default. Claude Code reads ",[66,2924,143],{},", and since recent versions also reads ",[66,2927,272],{}," when there is no ",[66,2930,143],{},". The simplest setup for a mixed team is one ",[66,2933,272],{}," as the source of truth, plus a one-line ",[66,2936,143],{}," that imports it:",[59,2939,2941],{"className":61,"code":2940,"language":63,"meta":64,"style":64},"echo \"@AGENTS.md\" > CLAUDE.md\n",[66,2942,2943],{"__ignoreMap":64},[69,2944,2945,2948,2951,2954],{"class":71,"line":72},[69,2946,2947],{"class":83},"echo",[69,2949,2950],{"class":79}," \"@AGENTS.md\"",[69,2952,2953],{"class":87}," >",[69,2955,2956],{"class":79}," CLAUDE.md\n",[21,2958,2960],{"id":2959},"enterprise-data-and-residency","Enterprise, data and residency",[11,2962,2963],{},"At the enterprise tier OpenAI lists SCIM, role-based access, audit logs through its Compliance API and no training on business data by default. On the Anthropic side, Team and Enterprise come with central admin and the managed settings shown above. The bigger difference for European companies is where inference runs.",[11,2965,2966,2968],{},[32,2967,40],{}," can be pointed at Amazon Bedrock, Google Cloud or Microsoft Foundry instead of Anthropic's API, so prompts and code are processed in the cloud region you pick. Model availability varies by provider and region.",[11,2970,2971,2973,2974,2977],{},[32,2972,2300],{}," can use a custom model provider. OpenAI's docs show an Azure OpenAI example in ",[66,2975,2976],{},"config.toml",", which sends requests to your own Azure resource:",[59,2979,2983],{"className":2980,"code":2981,"language":2982,"meta":64,"style":64},"language-toml shiki shiki-themes github-dark","# ~\u002F.codex\u002Fconfig.toml\nmodel_provider = \"azure\"\n\n[model_providers.azure]\nname = \"Azure\"\nbase_url = \"https:\u002F\u002FYOUR_PROJECT_NAME.openai.azure.com\u002Fopenai\"\nenv_key = \"AZURE_OPENAI_API_KEY\"\nquery_params = { api-version = \"2025-04-01-preview\" }\nwire_api = \"responses\"\n","toml",[66,2984,2985,2990,2998,3002,3018,3026,3034,3042,3053],{"__ignoreMap":64},[69,2986,2987],{"class":71,"line":72},[69,2988,2989],{"class":800},"# ~\u002F.codex\u002Fconfig.toml\n",[69,2991,2992,2995],{"class":71,"line":125},[69,2993,2994],{"class":479},"model_provider = ",[69,2996,2997],{"class":79},"\"azure\"\n",[69,2999,3000],{"class":71,"line":134},[69,3001,818],{"emptyLinePlaceholder":713},[69,3003,3004,3007,3010,3012,3015],{"class":71,"line":511},[69,3005,3006],{"class":479},"[",[69,3008,3009],{"class":75},"model_providers",[69,3011,149],{"class":479},[69,3013,3014],{"class":75},"azure",[69,3016,3017],{"class":479},"]\n",[69,3019,3020,3023],{"class":71,"line":517},[69,3021,3022],{"class":479},"name = ",[69,3024,3025],{"class":79},"\"Azure\"\n",[69,3027,3028,3031],{"class":71,"line":2440},[69,3029,3030],{"class":479},"base_url = ",[69,3032,3033],{"class":79},"\"https:\u002F\u002FYOUR_PROJECT_NAME.openai.azure.com\u002Fopenai\"\n",[69,3035,3036,3039],{"class":71,"line":2455},[69,3037,3038],{"class":479},"env_key = ",[69,3040,3041],{"class":79},"\"AZURE_OPENAI_API_KEY\"\n",[69,3043,3045,3048,3051],{"class":71,"line":3044},8,[69,3046,3047],{"class":479},"query_params = { api-version = ",[69,3049,3050],{"class":79},"\"2025-04-01-preview\"",[69,3052,508],{"class":479},[69,3054,3056,3059],{"class":71,"line":3055},9,[69,3057,3058],{"class":479},"wire_api = ",[69,3060,3061],{"class":79},"\"responses\"\n",[11,3063,3064],{},"Check which deployment type each model has in your region: the newest OpenAI models do not always launch in EU data zones.",[21,3066,3068],{"id":3067},"the-multi-model-argument","The multi-model argument",[11,3070,3071],{},"Most teams we talk to do not pick one. Some developers prefer Claude for long refactors, others prefer GPT for quick fixes, and the preference flips with each model release. Standardising on one vendor saves an invoice but locks you out of whichever model is ahead next quarter.",[11,3073,3074,3075,3077,3078,3081,3082,3085],{},"Running both is easy on the developer side, since the MCP servers, skills and ",[66,3076,272],{}," carry over. The hard part is governance: two sets of keys, two policy models, two audit trails, two places code leaves the building. That is the job of a gateway. ",[15,3079,3080],{"href":670},"Walma AI Hub"," runs in your own Azure tenant in an EU region and serves Claude, GPT and other models to Claude Code, Codex and Cursor through one endpoint, with per-team budgets, model policy, approved MCP servers and one log. Our ",[15,3083,3084],{"href":675},"MCP gateway guide"," explains the pattern in more depth.",[21,3087,3089],{"id":3088},"which-should-you-choose","Which should you choose?",[26,3091,3092,3098,3104],{},[29,3093,3094,3097],{},[32,3095,3096],{},"Choose Claude Code"," if you want fine-grained, centrally enforced permissions, routines and Slack-driven workflows, or if your developers prefer Claude models.",[29,3099,3100,3103],{},[32,3101,3102],{},"Choose Codex"," if your company already standardised on ChatGPT, you want a sandbox that is on from the first run, or your developers prefer GPT models.",[29,3105,3106,3109,3110,3112],{},[32,3107,3108],{},"Choose both"," if you have more than a handful of developers. Share one ",[66,3111,272],{},", keep skills to the open standard, and put policy and logging in a gateway rather than on each machine.",[679,3114,1525],{},{"title":64,"searchDepth":125,"depth":134,"links":3116},[3117,3118,3119,3120,3121,3122,3123,3124,3125],{"id":2286,"depth":125,"text":2287},{"id":2398,"depth":125,"text":2399},{"id":155,"depth":125,"text":156},{"id":2528,"depth":125,"text":2529},{"id":2614,"depth":125,"text":2615},{"id":2789,"depth":125,"text":2790},{"id":2959,"depth":125,"text":2960},{"id":3067,"depth":125,"text":3068},{"id":3088,"depth":125,"text":3089},"Claude Code vs OpenAI Codex compared on what matters to a team: surfaces and workflow, models, pricing, sandboxing and approvals, MCP, skills and AGENTS.md vs CLAUDE.md, and enterprise data residency. Plus why many teams end up running both.",[3128,3131,3134,3137,3140],{"q":3129,"a":3130},"What is the difference between Claude Code and Codex?","Both are agentic coding tools that read a repository, edit files and run commands. Claude Code is Anthropic's agent and runs Claude models (Opus, Sonnet, Haiku). Codex is OpenAI's agent and runs GPT models (GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna). The workflows are now very similar; the bigger differences are the models, the default safety posture and how each plugs into your cloud.",{"q":3132,"a":3133},"Is Codex cheaper than Claude Code?","The entry points are the same: both are included in a 20 US dollar per month individual plan, and both offer business seats from 20 dollars per user per month billed annually. Codex is also available on ChatGPT Free and Go. What you actually pay depends on usage limits, model choice and whether you bill per token through the API.",{"q":3135,"a":3136},"Can Claude Code read AGENTS.md?","Yes. Recent versions of Claude Code read AGENTS.md when the repository has no CLAUDE.md, and a setting lets it read both. You can also import AGENTS.md from a CLAUDE.md, so one instruction file can serve both agents.",{"q":3138,"a":3139},"Do Claude Code and Codex both support MCP and skills?","Yes. Both connect to MCP servers over stdio and HTTP, and both support skills in the open Agent Skills format (a folder with a SKILL.md file). Claude Code looks in .claude\u002Fskills, Codex in .agents\u002Fskills.",{"q":3141,"a":3142},"Can we run Claude Code and Codex with EU data residency?","Claude Code can run against Amazon Bedrock, Google Cloud or Microsoft Foundry instead of Anthropic's API, and Codex can point at an Azure OpenAI resource through a custom model provider. In both cases inference follows the cloud region you choose, subject to which models are deployed there. A gateway in your own EU tenant can serve both.",{},"\u002Fguides\u002Fclaude-code\u002Fclaude-code-vs-codex",{"title":2272,"description":3126},"guides\u002Fclaude-code\u002Fclaude-code-vs-codex","piJFFu4V2HArpes1pLF-OIZf_hlQdXCrW40RDiOPVho",{"id":3149,"title":3150,"author":6,"body":3151,"date":693,"description":4082,"extension":695,"faq":4083,"meta":4099,"navigation":713,"order":517,"path":4100,"readTime":715,"seo":4101,"stem":4102,"topic":718,"translationId":719,"updated":693,"__hash__":4103},"guides\u002Fguides\u002Fclaude-code\u002Fhow-to-use-claude-code.md","How to use Claude Code: a hands-on tutorial for daily work",{"type":8,"value":3152,"toc":4068},[3153,3162,3166,3169,3183,3186,3204,3208,3215,3306,3312,3321,3372,3379,3383,3386,3424,3427,3431,3437,3449,3499,3512,3516,3522,3603,3617,3621,3624,3657,3660,3663,3679,3733,3736,3754,3864,3877,3881,3887,3912,3920,3948,3952,3955,4007,4010,4014,4051,4055,4065],[11,3154,3155,3156,3158,3159,3161],{},"This tutorial starts where installation ends. If you have not installed the agent yet, follow ",[15,3157,867],{"href":147},", and if you want the big picture of what it is and what it costs, read the ",[15,3160,2283],{"href":17},". What follows is how to work with it day to day: the controls you touch every session, the files that make it better over time, and a full worked example. Everything here reflects Anthropic's Claude Code docs as of October 2026.",[21,3163,3165],{"id":3164},"your-first-session","Your first session",[11,3167,3168],{},"Start the agent from the root of a repository:",[59,3170,3171],{"className":61,"code":1248,"language":63,"meta":64,"style":64},[66,3172,3173,3179],{"__ignoreMap":64},[69,3174,3175,3177],{"class":71,"line":72},[69,3176,128],{"class":83},[69,3178,1257],{"class":79},[69,3180,3181],{"class":71,"line":125},[69,3182,137],{"class":75},[11,3184,3185],{},"Start with questions rather than code: \"give me an overview of this codebase\", \"how is authentication handled?\". You learn the repo and see how Claude navigates it at the same time.",[11,3187,3188,3189,265,3192,3195,3196,3199,3200,3203],{},"Reference files with ",[66,3190,3191],{},"@",[66,3193,3194],{},"@src\u002Futils\u002Fauth.js",") instead of describing where code lives, and paste screenshots of errors or mockups straight into the prompt. To pick a session back up later, ",[66,3197,3198],{},"claude --continue"," resumes the most recent conversation in that directory and ",[66,3201,3202],{},"claude --resume"," shows a list.",[21,3205,3207],{"id":3206},"permission-modes","Permission modes",[11,3209,3210,3211,3214],{},"A permission mode decides what Claude may do without asking. Press ",[32,3212,3213],{},"Shift+Tab"," to cycle through them during a session; the status bar shows the active one.",[158,3216,3217,3230],{},[161,3218,3219],{},[164,3220,3221,3224,3227],{},[167,3222,3223],{},"Mode",[167,3225,3226],{},"Runs without asking",[167,3228,3229],{},"Use it for",[174,3231,3232,3246,3258,3270,3282,3294],{},[164,3233,3234,3240,3243],{},[179,3235,3236,3237,2384],{},"Manual (",[66,3238,3239],{},"default",[179,3241,3242],{},"Reads only",[179,3244,3245],{},"Sensitive work, unfamiliar code",[164,3247,3248,3252,3255],{},[179,3249,3250],{},[66,3251,2680],{},[179,3253,3254],{},"Reads, file edits, common filesystem commands",[179,3256,3257],{},"Iterating on code you are reviewing",[164,3259,3260,3264,3267],{},[179,3261,3262],{},[66,3263,2683],{},[179,3265,3266],{},"Reads; no edits until you approve a plan",[179,3268,3269],{},"Exploring before changing anything",[164,3271,3272,3276,3279],{},[179,3273,3274],{},[66,3275,2687],{},[179,3277,3278],{},"Most actions, reviewed by a classifier model",[179,3280,3281],{},"Long tasks with fewer interruptions",[164,3283,3284,3288,3291],{},[179,3285,3286],{},[66,3287,2691],{},[179,3289,3290],{},"Only pre-approved tools, everything else denied",[179,3292,3293],{},"Locked-down CI and scripts",[164,3295,3296,3300,3303],{},[179,3297,3298],{},[66,3299,2694],{},[179,3301,3302],{},"Everything",[179,3304,3305],{},"Isolated containers and VMs only",[11,3307,3308,3309,149],{},"In current versions, auto mode is the starting mode for interactive terminal and VS Code sessions, provided you use a supported model and your organization has not turned it off. To approve every step yourself, start with ",[66,3310,3311],{},"claude --permission-mode manual",[11,3313,3314,3315,3317,3318,2712],{},"Modes set the baseline. On top of them, ",[66,3316,1360],{}," manages allow, ask and deny rules. Deny rules apply in every mode, so this is the place to block things like pushing to ",[66,3319,3320],{},"main",[59,3322,3324],{"className":470,"code":3323,"language":472,"meta":64,"style":64},"{\n  \"permissions\": {\n    \"allow\": [\"Bash(npm run test *)\", \"Bash(npm run lint)\"],\n    \"deny\": [\"Bash(git push origin main)\"]\n  }\n}\n",[66,3325,3326,3330,3336,3353,3364,3368],{"__ignoreMap":64},[69,3327,3328],{"class":71,"line":72},[69,3329,480],{"class":479},[69,3331,3332,3334],{"class":71,"line":125},[69,3333,2726],{"class":83},[69,3335,488],{"class":479},[69,3337,3338,3341,3343,3346,3348,3351],{"class":71,"line":134},[69,3339,3340],{"class":83},"    \"allow\"",[69,3342,2736],{"class":479},[69,3344,3345],{"class":79},"\"Bash(npm run test *)\"",[69,3347,269],{"class":479},[69,3349,3350],{"class":79},"\"Bash(npm run lint)\"",[69,3352,2747],{"class":479},[69,3354,3355,3357,3359,3362],{"class":71,"line":511},[69,3356,2733],{"class":83},[69,3358,2736],{"class":479},[69,3360,3361],{"class":79},"\"Bash(git push origin main)\"",[69,3363,3017],{"class":479},[69,3365,3366],{"class":71,"line":517},[69,3367,514],{"class":479},[69,3369,3370],{"class":71,"line":2440},[69,3371,520],{"class":479},[11,3373,3374,3375,3378],{},"That block goes in ",[66,3376,3377],{},".claude\u002Fsettings.json"," and is shared with the team through git.",[21,3380,3382],{"id":3381},"plan-mode-explore-plan-then-code","Plan mode: explore, plan, then code",[11,3384,3385],{},"For anything that touches several files, separate thinking from typing. Anthropic's recommended loop has four steps:",[1745,3387,3388,3402,3412,3418],{},[29,3389,3390,3393,3394,3397,3398,3401],{},[32,3391,3392],{},"Explore"," in plan mode (Shift+Tab until ",[66,3395,3396],{},"plan mode on",", or ",[66,3399,3400],{},"\u002Fplan"," in front of one prompt). Claude reads and answers without editing.",[29,3403,3404,3407,3408,3411],{},[32,3405,3406],{},"Plan."," Ask what needs to change and for a step-by-step plan. Press ",[32,3409,3410],{},"Ctrl+G"," to open the plan in your editor and fix it directly.",[29,3413,3414,3417],{},[32,3415,3416],{},"Implement."," Approve the plan; the session leaves plan mode and Claude starts editing, checking its work against the plan.",[29,3419,3420,3423],{},[32,3421,3422],{},"Commit."," Ask for a descriptive commit and a pull request.",[11,3425,3426],{},"Skip the plan when you could describe the diff in one sentence.",[21,3428,3430],{"id":3429},"claudemd-and-init","CLAUDE.md and \u002Finit",[11,3432,3433,3434,3436],{},"Every session starts with an empty context window. ",[66,3435,143],{}," is how knowledge survives between sessions: Claude loads it at the start of every conversation.",[11,3438,3439,3440,3442,3443,3445,3446,3448],{},"Run ",[66,3441,919],{}," once per repository. It analyzes the codebase and drafts a ",[66,3444,143],{}," with build commands, test instructions and conventions it finds. If one already exists, it suggests improvements instead of overwriting. Then edit by hand. Keep commands Claude cannot guess, style rules that differ from defaults, PR conventions and gotchas. Leave out anything Claude can learn by reading the code, and keep each file under about 200 lines. A bloated ",[66,3447,143],{}," gets followed less, not more. Where the files live:",[158,3450,3451,3461],{},[161,3452,3453],{},[164,3454,3455,3458],{},[167,3456,3457],{},"File",[167,3459,3460],{},"Scope",[174,3462,3463,3476,3489],{},[164,3464,3465,3473],{},[179,3466,3467,2511,3470],{},[66,3468,3469],{},".\u002FCLAUDE.md",[66,3471,3472],{},".\u002F.claude\u002FCLAUDE.md",[179,3474,3475],{},"The project, committed for the team",[164,3477,3478,3483],{},[179,3479,3480],{},[66,3481,3482],{},".\u002FCLAUDE.local.md",[179,3484,3485,3486,2384],{},"Your personal notes for this project (add to ",[66,3487,3488],{},".gitignore",[164,3490,3491,3496],{},[179,3492,3493],{},[66,3494,3495],{},"~\u002F.claude\u002FCLAUDE.md",[179,3497,3498],{},"You, across all projects",[11,3500,3501,3504,3505,3508,3509,3511],{},[66,3502,3503],{},"\u002Fmemory"," opens these files for editing, and ",[66,3506,3507],{},"@path\u002Fto\u002Ffile"," inside a ",[66,3510,143],{}," imports another file.",[21,3513,3515],{"id":3514},"the-slash-commands-you-will-actually-use","The slash commands you will actually use",[11,3517,3518,3519,3521],{},"Type ",[66,3520,1187],{}," to see everything. These are the ones that come up daily:",[158,3523,3524,3533],{},[161,3525,3526],{},[164,3527,3528,3530],{},[167,3529,1319],{},[167,3531,3532],{},"What it does",[174,3534,3535,3544,3554,3564,3574,3584,3593],{},[164,3536,3537,3541],{},[179,3538,3539],{},[66,3540,1370],{},[179,3542,3543],{},"New conversation, empty context",[164,3545,3546,3551],{},[179,3547,3548],{},[66,3549,3550],{},"\u002Fcompact [focus]",[179,3552,3553],{},"Summarize the conversation to free space",[164,3555,3556,3561],{},[179,3557,3558],{},[66,3559,3560],{},"\u002Fcontext",[179,3562,3563],{},"Show what is filling the context window",[164,3565,3566,3571],{},[179,3567,3568],{},[66,3569,3570],{},"\u002Frewind",[179,3572,3573],{},"Restore conversation and\u002For code to an earlier checkpoint (also Esc Esc)",[164,3575,3576,3581],{},[179,3577,3578],{},[66,3579,3580],{},"\u002Fplan [task]",[179,3582,3583],{},"Enter plan mode, optionally with a task",[164,3585,3586,3590],{},[179,3587,3588],{},[66,3589,938],{},[179,3591,3592],{},"Manage MCP servers and their OAuth logins",[164,3594,3595,3600],{},[179,3596,3597],{},[66,3598,3599],{},"\u002Fcode-review",[179,3601,3602],{},"Review the current diff for bugs in a fresh subagent",[11,3604,3605,3606,3609,3610,3613,3614,149],{},"Your own commands are now skills (more on that below): ",[66,3607,3608],{},".claude\u002Fcommands\u002Fdeploy.md"," still works and creates ",[66,3611,3612],{},"\u002Fdeploy",", exactly like ",[66,3615,3616],{},".claude\u002Fskills\u002Fdeploy\u002FSKILL.md",[21,3618,3620],{"id":3619},"managing-context","Managing context",[11,3622,3623],{},"The context window is the resource that matters most. Every file read and every command output lands in it, and quality drops as it fills. The habits that keep sessions sharp:",[26,3625,3626,3634,3643,3651],{},[29,3627,3628,3633],{},[32,3629,3630,3632],{},[66,3631,1370],{}," between unrelated tasks."," Mixing three topics in one session is the most common failure.",[29,3635,3636,3639,3640,3642],{},[32,3637,3638],{},"Correct early."," Esc stops Claude mid-action with context intact. After two failed corrections, ",[66,3641,1370],{}," and write a better prompt.",[29,3644,3645,3650],{},[32,3646,3647],{},[66,3648,3649],{},"\u002Fcompact Focus on the API changes"," when you want to keep going but trim history.",[29,3652,3653,3656],{},[32,3654,3655],{},"Rewind instead of arguing."," Checkpoints are taken on every prompt, so you can let Claude try something risky and roll back. They only track edits made through Claude's own file tools, not Bash, so git remains your real safety net.",[21,3658,284],{"id":3659},"subagents",[11,3661,3662],{},"A subagent runs in its own context window and reports back a summary, so exploration does not fill your main session. Built-in ones include Explore (read-only search) and Plan. The simplest use is to ask: \"use a subagent to investigate how our auth system handles token refresh\".",[11,3664,3665,3666,3669,3670,3673,3674,261,3676,3678],{},"Custom subagents are Markdown files in ",[66,3667,3668],{},".claude\u002Fagents\u002F"," (project) or ",[66,3671,3672],{},"~\u002F.claude\u002Fagents\u002F"," (personal). Only ",[66,3675,2881],{},[66,3677,2885],{}," are required in the frontmatter; the body is the agent's instructions:",[59,3680,3684],{"className":3681,"code":3682,"language":3683,"meta":64,"style":64},"language-yaml shiki shiki-themes github-dark","---\nname: security-reviewer\ndescription: Reviews code for security vulnerabilities\ntools: Read, Grep, Glob, Bash\nmodel: opus\n---\n","yaml",[66,3685,3686,3691,3701,3710,3720,3729],{"__ignoreMap":64},[69,3687,3688],{"class":71,"line":72},[69,3689,3690],{"class":75},"---\n",[69,3692,3693,3696,3698],{"class":71,"line":125},[69,3694,2881],{"class":3695},"s4JwU",[69,3697,502],{"class":479},[69,3699,3700],{"class":79},"security-reviewer\n",[69,3702,3703,3705,3707],{"class":71,"line":134},[69,3704,2885],{"class":3695},[69,3706,502],{"class":479},[69,3708,3709],{"class":79},"Reviews code for security vulnerabilities\n",[69,3711,3712,3715,3717],{"class":71,"line":511},[69,3713,3714],{"class":3695},"tools",[69,3716,502],{"class":479},[69,3718,3719],{"class":79},"Read, Grep, Glob, Bash\n",[69,3721,3722,3724,3726],{"class":71,"line":517},[69,3723,2518],{"class":3695},[69,3725,502],{"class":479},[69,3727,3728],{"class":79},"opus\n",[69,3730,3731],{"class":71,"line":2440},[69,3732,3690],{"class":75},[21,3734,260],{"id":3735},"hooks",[11,3737,3738,3740,3741,269,3744,269,3747,2511,3750,3753],{},[66,3739,143],{}," is advice. A hook is a guarantee: a shell command Claude Code runs at a fixed point in its lifecycle, such as ",[66,3742,3743],{},"PreToolUse",[66,3745,3746],{},"PostToolUse",[66,3748,3749],{},"Notification",[66,3751,3752],{},"Stop",". This example from Anthropic's docs runs Prettier on every file Claude edits:",[59,3755,3757],{"className":470,"code":3756,"language":472,"meta":64,"style":64},"{\n  \"hooks\": {\n    \"PostToolUse\": [\n      {\n        \"matcher\": \"Edit|Write\",\n        \"hooks\": [\n          {\n            \"type\": \"command\",\n            \"command\": \"jq -r '.tool_input.file_path' | xargs npx prettier --write\"\n          }\n        ]\n      }\n    ]\n  }\n}\n",[66,3758,3759,3763,3770,3778,3783,3796,3803,3808,3820,3830,3836,3842,3848,3854,3859],{"__ignoreMap":64},[69,3760,3761],{"class":71,"line":72},[69,3762,480],{"class":479},[69,3764,3765,3768],{"class":71,"line":125},[69,3766,3767],{"class":83},"  \"hooks\"",[69,3769,488],{"class":479},[69,3771,3772,3775],{"class":71,"line":134},[69,3773,3774],{"class":83},"    \"PostToolUse\"",[69,3776,3777],{"class":479},": [\n",[69,3779,3780],{"class":71,"line":511},[69,3781,3782],{"class":479},"      {\n",[69,3784,3785,3788,3790,3793],{"class":71,"line":517},[69,3786,3787],{"class":83},"        \"matcher\"",[69,3789,502],{"class":479},[69,3791,3792],{"class":79},"\"Edit|Write\"",[69,3794,3795],{"class":479},",\n",[69,3797,3798,3801],{"class":71,"line":2440},[69,3799,3800],{"class":83},"        \"hooks\"",[69,3802,3777],{"class":479},[69,3804,3805],{"class":71,"line":2455},[69,3806,3807],{"class":479},"          {\n",[69,3809,3810,3813,3815,3818],{"class":71,"line":3044},[69,3811,3812],{"class":83},"            \"type\"",[69,3814,502],{"class":479},[69,3816,3817],{"class":79},"\"command\"",[69,3819,3795],{"class":479},[69,3821,3822,3825,3827],{"class":71,"line":3055},[69,3823,3824],{"class":83},"            \"command\"",[69,3826,502],{"class":479},[69,3828,3829],{"class":79},"\"jq -r '.tool_input.file_path' | xargs npx prettier --write\"\n",[69,3831,3833],{"class":71,"line":3832},10,[69,3834,3835],{"class":479},"          }\n",[69,3837,3839],{"class":71,"line":3838},11,[69,3840,3841],{"class":479},"        ]\n",[69,3843,3845],{"class":71,"line":3844},12,[69,3846,3847],{"class":479},"      }\n",[69,3849,3851],{"class":71,"line":3850},13,[69,3852,3853],{"class":479},"    ]\n",[69,3855,3857],{"class":71,"line":3856},14,[69,3858,514],{"class":479},[69,3860,3862],{"class":71,"line":3861},15,[69,3863,520],{"class":479},[11,3865,3866,3867,3869,3870,3873,3874,3876],{},"Put it in ",[66,3868,3377],{}," and check it with ",[66,3871,3872],{},"\u002Fhooks",". If Claude keeps breaking a rule in ",[66,3875,143],{},", turn it into a hook.",[21,3878,3880],{"id":3879},"mcp-and-skills-in-brief","MCP and skills in brief",[11,3882,3883,3886],{},[32,3884,3885],{},"MCP servers"," connect Claude Code to external systems such as issue trackers, databases and design tools:",[59,3888,3890],{"className":61,"code":3889,"language":63,"meta":64,"style":64},"claude mcp add --transport http notion https:\u002F\u002Fmcp.notion.com\u002Fmcp\n",[66,3891,3892],{"__ignoreMap":64},[69,3893,3894,3896,3898,3900,3903,3906,3909],{"class":71,"line":72},[69,3895,658],{"class":75},[69,3897,2810],{"class":79},[69,3899,2813],{"class":79},[69,3901,3902],{"class":83}," --transport",[69,3904,3905],{"class":79}," http",[69,3907,3908],{"class":79}," notion",[69,3910,3911],{"class":79}," https:\u002F\u002Fmcp.notion.com\u002Fmcp\n",[11,3913,861,3914,3916,3917,149],{},[66,3915,938],{}," inside a session to log in. The full walkthrough is in ",[15,3918,3919],{"href":467},"adding an MCP server to Claude Code",[11,3921,3922,3925,3926,2519,3928,3931,3932,3934,3935,3937,3938,3941,3942,3945,3946,149],{},[32,3923,3924],{},"Skills"," are folders with a ",[66,3927,529],{},[66,3929,3930],{},".claude\u002Fskills\u002F\u003Cname>\u002F"," (or ",[66,3933,536],{}," for personal ones). Claude loads one automatically when its ",[66,3936,2885],{}," matches the task, or you run it as ",[66,3939,3940],{},"\u002Fname",". Add ",[66,3943,3944],{},"disable-model-invocation: true"," for workflows with side effects, like deploys, that should only run when you type the command. See ",[15,3947,954],{"href":540},[21,3949,3951],{"id":3950},"worked-example-from-bug-report-to-pull-request","Worked example: from bug report to pull request",[11,3953,3954],{},"A realistic flow for a bug report that says login fails after a session timeout:",[1745,3956,3957,3966,3972,3978,3984,3993],{},[29,3958,3959,3962,3963,3965],{},[32,3960,3961],{},"Start clean."," ",[66,3964,1370],{},", or a new session. In plan mode, prompt: \"users report that login fails after session timeout. check the auth flow in src\u002Fauth\u002F, especially token refresh. explain the likely cause\".",[29,3967,3968,3971],{},[32,3969,3970],{},"Reproduce first."," \"Write a failing test that reproduces the issue. Run it and show me the output.\" A failing test is the check Claude will iterate against.",[29,3973,3974,3977],{},[32,3975,3976],{},"Plan and approve."," Ask for a minimal fix plan, adjust it with Ctrl+G if needed, approve.",[29,3979,3980,3983],{},[32,3981,3982],{},"Fix and verify."," \"Implement the fix, run the auth tests and the linter, fix any failures. Address the root cause, do not suppress the error.\" Ask Claude to show the test output rather than just claim success.",[29,3985,3986,3989,3990,3992],{},[32,3987,3988],{},"Review."," Run ",[66,3991,3599],{}," so a fresh subagent looks at the diff without the reasoning that produced it.",[29,3994,3995,3998,3999,4002,4003,4006],{},[32,3996,3997],{},"Ship."," \"Commit with a descriptive message and open a PR.\" With ",[66,4000,4001],{},"gh"," installed and authenticated, Claude creates the PR and links the session to it; ",[66,4004,4005],{},"claude --from-pr \u003Cnumber>"," finds that session again later.",[11,4008,4009],{},"The key move is step 2: give Claude a check it can run, and it closes the loop itself instead of waiting for you to spot mistakes.",[21,4011,4013],{"id":4012},"tips-that-actually-help","Tips that actually help",[26,4015,4016,4022,4028,4038],{},[29,4017,4018,4021],{},[32,4019,4020],{},"Be specific."," Name files, the scenario and what \"done\" looks like. \"Add tests for foo.py\" is weaker than \"write a test for foo.py covering the logged-out case, avoid mocks\".",[29,4023,4024,4027],{},[32,4025,4026],{},"Point to existing patterns"," in the repo rather than describing them.",[29,4029,4030,4033,4034,4037],{},[32,4031,4032],{},"Run parallel work in worktrees"," with ",[66,4035,4036],{},"claude --worktree feature-auth"," so edits do not collide.",[29,4039,4040,4033,4043,4046,4047,4050],{},[32,4041,4042],{},"Name sessions",[66,4044,4045],{},"\u002Frename"," so ",[66,4048,4049],{},"\u002Fresume"," stays useful.",[21,4052,4054],{"id":4053},"rolling-it-out-to-a-team","Rolling it out to a team",[11,4056,4057,4058,4061,4062,4064],{},"For a team, the open questions are where traffic goes, which MCP servers are allowed and who can see what was done. Claude Code can run against an LLM gateway instead of Anthropic directly, which puts model access, MCP policy and logging in one place. Walma AI Hub is that gateway, running in the customer's own Azure tenant in an EU region; see ",[15,4059,4060],{"href":670},"the AI Hub"," or the ",[15,4063,3084],{"href":675}," for how the controls fit together.",[679,4066,4067],{},"html pre.shiki code .sDLfK, html code.shiki .sDLfK{--shiki-default:#79B8FF}html pre.shiki code .sU2Wk, html code.shiki .sU2Wk{--shiki-default:#9ECBFF}html pre.shiki code .svObZ, html code.shiki .svObZ{--shiki-default:#B392F0}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html pre.shiki code .s95oV, html code.shiki .s95oV{--shiki-default:#E1E4E8}html pre.shiki code .s4JwU, html code.shiki .s4JwU{--shiki-default:#85E89D}",{"title":64,"searchDepth":125,"depth":134,"links":4069},[4070,4071,4072,4073,4074,4075,4076,4077,4078,4079,4080,4081],{"id":3164,"depth":125,"text":3165},{"id":3206,"depth":125,"text":3207},{"id":3381,"depth":125,"text":3382},{"id":3429,"depth":125,"text":3430},{"id":3514,"depth":125,"text":3515},{"id":3619,"depth":125,"text":3620},{"id":3659,"depth":125,"text":284},{"id":3735,"depth":125,"text":260},{"id":3879,"depth":125,"text":3880},{"id":3950,"depth":125,"text":3951},{"id":4012,"depth":125,"text":4013},{"id":4053,"depth":125,"text":4054},"A practical Claude Code tutorial for after installation: your first session, permission modes, plan mode, CLAUDE.md and \u002Finit, the slash commands that matter, context management, subagents, hooks, MCP and skills, plus a worked example from bug to tested pull request.",[4084,4087,4090,4093,4096],{"q":4085,"a":4086},"How do I start using Claude Code?","Open a terminal in your project folder and run claude. Ask for an overview of the codebase first, then run \u002Finit to generate a CLAUDE.md with your build and test commands. From there, describe tasks in plain language and let Claude read, edit and run commands, reviewing as it goes.",{"q":4088,"a":4089},"What are the permission modes in Claude Code?","Manual (config value default) asks before edits and commands. acceptEdits auto-approves file edits and common filesystem commands. plan lets Claude read and propose a plan without editing. auto lets a classifier model review actions instead of you. dontAsk only runs pre-approved tools, and bypassPermissions skips checks and is meant for isolated containers. Press Shift+Tab to cycle modes during a session.",{"q":4091,"a":4092},"How do I use plan mode in Claude Code?","Press Shift+Tab until the status bar shows plan mode on, prefix a single prompt with \u002Fplan, or start with claude --permission-mode plan. Claude explores and writes a plan without editing your source. Press Ctrl+G to edit the plan in your text editor, then approve it to let Claude start implementing.",{"q":4094,"a":4095},"What is the difference between \u002Fclear and \u002Fcompact?","\u002Fclear starts a new conversation with an empty context, which is what you want between unrelated tasks. \u002Fcompact keeps the same conversation but summarizes it to free space, and accepts focus instructions such as \u002Fcompact Focus on the API changes.",{"q":4097,"a":4098},"Do custom slash commands still exist in Claude Code?","Yes, but they have been merged into skills. A file at .claude\u002Fcommands\u002Fdeploy.md and a skill at .claude\u002Fskills\u002Fdeploy\u002FSKILL.md both create \u002Fdeploy. Existing command files keep working; skills add supporting files and frontmatter that controls whether Claude can invoke them on its own.",{},"\u002Fguides\u002Fclaude-code\u002Fhow-to-use-claude-code",{"title":3150,"description":4082},"guides\u002Fclaude-code\u002Fhow-to-use-claude-code","gMQmoqbyVHoKoOd1_cg0XiiqSCuHEC29LnPtD7AS0EE",[],1791200328813]