[{"data":1,"prerenderedAt":880},["ShallowReactive",2],{"blog-post-en-claude-code-agent-teams-guide":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"tags":11,"body":14,"_type":874,"_id":875,"_source":876,"_file":877,"_stem":878,"_extension":879},"\u002Fen\u002Fblog\u002Fclaude-code-agent-teams-guide","blog",false,"","Claude Code Agent Teams: A Practical Introduction","What Claude Code Agent Teams is, how the lead, mailbox, and shared task list work, what it's good for, and how to enable it — with honest caveats.","2026-07-18",[12,13],"claude-code","agent-frameworks",{"type":15,"children":16,"toc":860},"root",[17,44,49,56,75,80,105,119,140,146,151,244,251,265,271,283,288,294,313,319,333,357,375,381,386,436,447,453,474,486,492,505,580,585,617,622,713,718,724,729,789,795,837,841,854],{"type":18,"tag":19,"props":20,"children":21},"element","p",{},[22,25,34,36,42],{"type":23,"value":24},"text","When Anthropic shipped ",{"type":18,"tag":26,"props":27,"children":31},"a",{"href":28,"rel":29},"https:\u002F\u002Fwww.anthropic.com\u002Fnews\u002Fclaude-opus-4-6",[30],"nofollow",[32],{"type":23,"value":33},"Claude Opus 4.6",{"type":23,"value":35}," in early February 2026, the model got the headlines. But the feature that matters most for anyone interested in AI teams was tucked into Claude Code: ",{"type":18,"tag":37,"props":38,"children":39},"strong",{},[40],{"type":23,"value":41},"agent teams",{"type":23,"value":43},", a research-preview capability that lets one Claude Code session spawn and coordinate several other Claude Code sessions working in parallel. It is the closest thing yet to watching a small AI engineering team form, divide up work, argue about findings, and report back — all inside your terminal.",{"type":18,"tag":19,"props":45,"children":46},{},[47],{"type":23,"value":48},"This is a practical introduction: what agent teams actually is, how the pieces fit together, where it genuinely helps, where it doesn't, and how to turn it on.",{"type":18,"tag":50,"props":51,"children":53},"h2",{"id":52},"what-agent-teams-is-and-isnt",[54],{"type":23,"value":55},"What agent teams is (and isn't)",{"type":18,"tag":19,"props":57,"children":58},{},[59,61,66,68,73],{"type":23,"value":60},"Agent teams is experimental multi-agent orchestration built directly into Claude Code. One session — the one you're typing into — becomes the ",{"type":18,"tag":37,"props":62,"children":63},{},[64],{"type":23,"value":65},"team lead",{"type":23,"value":67},". On your instruction, it spawns ",{"type":18,"tag":37,"props":69,"children":70},{},[71],{"type":23,"value":72},"teammates",{"type":23,"value":74},": each one a full, independent Claude Code instance with its own context window, its own tool access, and its own view of your project.",{"type":18,"tag":19,"props":76,"children":77},{},[78],{"type":23,"value":79},"Two things distinguish this from earlier \"spawn a helper\" patterns:",{"type":18,"tag":81,"props":82,"children":83},"ol",{},[84,95],{"type":18,"tag":85,"props":86,"children":87},"li",{},[88,93],{"type":18,"tag":37,"props":89,"children":90},{},[91],{"type":23,"value":92},"Teammates talk to each other",{"type":23,"value":94},", not just to the lead. Communication is peer-to-peer.",{"type":18,"tag":85,"props":96,"children":97},{},[98,103],{"type":18,"tag":37,"props":99,"children":100},{},[101],{"type":23,"value":102},"Work is coordinated through a shared task list",{"type":23,"value":104}," that teammates can claim from on their own, rather than the lead micromanaging every step.",{"type":18,"tag":19,"props":106,"children":107},{},[108,110,117],{"type":23,"value":109},"The feature is explicitly experimental. It's disabled by default, hidden behind an environment variable, and the ",{"type":18,"tag":26,"props":111,"children":114},{"href":112,"rel":113},"https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fagent-teams",[30],[115],{"type":23,"value":116},"official documentation",{"type":23,"value":118}," lists known limitations up front. Treat it as a preview of a working style, not a finished product.",{"type":18,"tag":19,"props":120,"children":121},{},[122,124,130,132,138],{"type":23,"value":123},"It's also worth being clear about what it is ",{"type":18,"tag":125,"props":126,"children":127},"em",{},[128],{"type":23,"value":129},"not",{"type":23,"value":131},": agent teams is an interactive development-workflow tool. It runs inside your terminal session, expects you to watch and steer, and its teams live and die with the session. It is not a production agent framework — there's no deployment story, no durable orchestration, no API for running teams as a service. If you're looking for that, see our ",{"type":18,"tag":26,"props":133,"children":135},{"href":134},"\u002Fblog\u002Fbest-multi-agent-frameworks-2026",[136],{"type":23,"value":137},"comparison of multi-agent frameworks",{"type":23,"value":139},".",{"type":18,"tag":50,"props":141,"children":143},{"id":142},"how-it-works-four-moving-parts",[144],{"type":23,"value":145},"How it works: four moving parts",{"type":18,"tag":19,"props":147,"children":148},{},[149],{"type":23,"value":150},"An agent team has four components, per the docs:",{"type":18,"tag":152,"props":153,"children":154},"table",{},[155,175],{"type":18,"tag":156,"props":157,"children":158},"thead",{},[159],{"type":18,"tag":160,"props":161,"children":162},"tr",{},[163,170],{"type":18,"tag":164,"props":165,"children":167},"th",{"align":166},"left",[168],{"type":23,"value":169},"Component",{"type":18,"tag":164,"props":171,"children":172},{"align":166},[173],{"type":23,"value":174},"What it does",{"type":18,"tag":176,"props":177,"children":178},"tbody",{},[179,196,212,228],{"type":18,"tag":160,"props":180,"children":181},{},[182,191],{"type":18,"tag":183,"props":184,"children":185},"td",{"align":166},[186],{"type":18,"tag":37,"props":187,"children":188},{},[189],{"type":23,"value":190},"Team lead",{"type":18,"tag":183,"props":192,"children":193},{"align":166},[194],{"type":23,"value":195},"Your main session. Spawns teammates, creates tasks, synthesizes results",{"type":18,"tag":160,"props":197,"children":198},{},[199,207],{"type":18,"tag":183,"props":200,"children":201},{"align":166},[202],{"type":18,"tag":37,"props":203,"children":204},{},[205],{"type":23,"value":206},"Teammates",{"type":18,"tag":183,"props":208,"children":209},{"align":166},[210],{"type":23,"value":211},"Independent Claude Code instances, each with its own context window",{"type":18,"tag":160,"props":213,"children":214},{},[215,223],{"type":18,"tag":183,"props":216,"children":217},{"align":166},[218],{"type":18,"tag":37,"props":219,"children":220},{},[221],{"type":23,"value":222},"Task list",{"type":18,"tag":183,"props":224,"children":225},{"align":166},[226],{"type":23,"value":227},"Shared work items that teammates claim and complete",{"type":18,"tag":160,"props":229,"children":230},{},[231,239],{"type":18,"tag":183,"props":232,"children":233},{"align":166},[234],{"type":18,"tag":37,"props":235,"children":236},{},[237],{"type":23,"value":238},"Mailbox",{"type":18,"tag":183,"props":240,"children":241},{"align":166},[242],{"type":23,"value":243},"Peer-to-peer messaging between any two agents",{"type":18,"tag":245,"props":246,"children":248},"h3",{"id":247},"the-mailbox",[249],{"type":23,"value":250},"The mailbox",{"type":18,"tag":19,"props":252,"children":253},{},[254,256,263],{"type":23,"value":255},"Each agent gets a mailbox — implemented as a JSON file under ",{"type":18,"tag":257,"props":258,"children":260},"code",{"className":259},[],[261],{"type":23,"value":262},"~\u002F.claude\u002Fteams\u002F{team-name}\u002Finboxes\u002F",{"type":23,"value":264}," — and any teammate can message any other teammate by name. This sounds like plumbing, but it's the architecturally interesting part: findings, objections, and requests flow directly between agents instead of being routed through the lead. The lead doesn't have to poll for updates either; messages are delivered automatically, and idle teammates notify the lead when they finish.",{"type":18,"tag":245,"props":266,"children":268},{"id":267},"the-shared-task-list",[269],{"type":23,"value":270},"The shared task list",{"type":18,"tag":19,"props":272,"children":273},{},[274,276,281],{"type":23,"value":275},"The lead breaks work into tasks with three states (pending, in progress, completed) and optional dependencies — a task that depends on unfinished work can't be claimed until the blocker completes. Teammates can be assigned tasks explicitly, or they can ",{"type":18,"tag":37,"props":277,"children":278},{},[279],{"type":23,"value":280},"self-claim",{"type":23,"value":282},": finish one task, pick up the next unassigned, unblocked one. Claiming uses file locking so two teammates can't grab the same task in a race.",{"type":18,"tag":19,"props":284,"children":285},{},[286],{"type":23,"value":287},"This is what makes the system feel like a team rather than a dispatcher with workers. You can give the lead fifteen independent tasks and three teammates, and the work distributes itself.",{"type":18,"tag":245,"props":289,"children":291},{"id":290},"what-teammates-know",[292],{"type":23,"value":293},"What teammates know",{"type":18,"tag":19,"props":295,"children":296},{},[297,299,305,307,311],{"type":23,"value":298},"A freshly spawned teammate loads the same project context a normal session would — ",{"type":18,"tag":257,"props":300,"children":302},{"className":301},[],[303],{"type":23,"value":304},"CLAUDE.md",{"type":23,"value":306},", MCP servers, skills — plus the spawn prompt the lead wrote for it. What it does ",{"type":18,"tag":125,"props":308,"children":309},{},[310],{"type":23,"value":129},{"type":23,"value":312}," get is the lead's conversation history. The practical consequence: put task-specific context into the spawn prompt, because the teammate wasn't in the room when you explained the problem.",{"type":18,"tag":50,"props":314,"children":316},{"id":315},"agent-teams-vs-subagents",[317],{"type":23,"value":318},"Agent teams vs. subagents",{"type":18,"tag":19,"props":320,"children":321},{},[322,324,331],{"type":23,"value":323},"Claude Code has had ",{"type":18,"tag":26,"props":325,"children":328},{"href":326,"rel":327},"https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fsub-agents",[30],[329],{"type":23,"value":330},"subagents",{"type":23,"value":332}," for a while, and the distinction matters:",{"type":18,"tag":334,"props":335,"children":336},"ul",{},[337,347],{"type":18,"tag":85,"props":338,"children":339},{},[340,345],{"type":18,"tag":37,"props":341,"children":342},{},[343],{"type":23,"value":344},"Subagents",{"type":23,"value":346}," run within your session, do a focused job, and report results back to the caller. They never talk to each other. Cheaper, simpler, right answer for \"go look this up and tell me what you found.\"",{"type":18,"tag":85,"props":348,"children":349},{},[350,355],{"type":18,"tag":37,"props":351,"children":352},{},[353],{"type":23,"value":354},"Agent teams",{"type":23,"value":356}," are fully independent sessions that share a task list and message each other. More expensive, more capable, right answer for work that benefits from discussion and parallel ownership.",{"type":18,"tag":19,"props":358,"children":359},{},[360,362,367,369,374],{"type":23,"value":361},"A useful rule from the docs: use subagents when only the ",{"type":18,"tag":125,"props":363,"children":364},{},[365],{"type":23,"value":366},"result",{"type":23,"value":368}," matters; use a team when the workers need to ",{"type":18,"tag":125,"props":370,"children":371},{},[372],{"type":23,"value":373},"coordinate",{"type":23,"value":139},{"type":18,"tag":50,"props":376,"children":378},{"id":377},"what-its-good-for",[379],{"type":23,"value":380},"What it's good for",{"type":18,"tag":19,"props":382,"children":383},{},[384],{"type":23,"value":385},"The pattern shines where parallel exploration adds real value:",{"type":18,"tag":334,"props":387,"children":388},{},[389,399,416,426],{"type":18,"tag":85,"props":390,"children":391},{},[392,397],{"type":18,"tag":37,"props":393,"children":394},{},[395],{"type":23,"value":396},"Parallel code review.",{"type":23,"value":398}," Spawn three reviewers with different lenses — security, performance, test coverage — over the same PR. Each applies one filter thoroughly instead of one agent skimming for everything at once.",{"type":18,"tag":85,"props":400,"children":401},{},[402,407,409,414],{"type":18,"tag":37,"props":403,"children":404},{},[405],{"type":23,"value":406},"Debugging with competing hypotheses.",{"type":23,"value":408}," The documentation's most interesting example: spawn several investigators, give each a different theory, and tell them to actively try to ",{"type":18,"tag":125,"props":410,"children":411},{},[412],{"type":23,"value":413},"disprove each other",{"type":23,"value":415}," like a scientific debate. Sequential investigation anchors on the first plausible explanation; adversarial parallel investigation is more likely to surface the real root cause.",{"type":18,"tag":85,"props":417,"children":418},{},[419,424],{"type":18,"tag":37,"props":420,"children":421},{},[422],{"type":23,"value":423},"New modules and features.",{"type":23,"value":425}," Work that splits along clean boundaries — one teammate on the API, one on the frontend, one on tests — where nobody edits the same file.",{"type":18,"tag":85,"props":427,"children":428},{},[429,434],{"type":18,"tag":37,"props":430,"children":431},{},[432],{"type":23,"value":433},"Research and exploration.",{"type":23,"value":435}," Have teammates evaluate a library, an architecture question, or a migration path from different angles, then synthesize.",{"type":18,"tag":19,"props":437,"children":438},{},[439,441,445],{"type":23,"value":440},"And what it's ",{"type":18,"tag":125,"props":442,"children":443},{},[444],{"type":23,"value":129},{"type":23,"value":446}," good for: sequential work, tasks with heavy interdependencies, and anything where multiple agents would edit the same files (two teammates writing one file ends in overwrites). For those, a single session or subagents are both cheaper and better.",{"type":18,"tag":50,"props":448,"children":450},{"id":449},"a-real-data-point-claude-code-review",[451],{"type":23,"value":452},"A real data point: Claude Code Review",{"type":18,"tag":19,"props":454,"children":455},{},[456,458,463,465,472],{"type":23,"value":457},"The clearest evidence that this pattern does real work came in March 2026, when Anthropic launched ",{"type":18,"tag":37,"props":459,"children":460},{},[461],{"type":23,"value":462},"Claude Code Review",{"type":23,"value":464},", an automated PR reviewer ",{"type":18,"tag":26,"props":466,"children":469},{"href":467,"rel":468},"https:\u002F\u002Fwww.vktr.com\u002Fai-news\u002Fanthropic-launches-multi-agent-code-review-for-claude-code\u002F",[30],[470],{"type":23,"value":471},"built on a multi-agent architecture",{"type":23,"value":473},": parallel specialized reviewer agents examine each pull request, followed by a verification pass that filters false positives before findings are posted as inline comments.",{"type":18,"tag":19,"props":475,"children":476},{},[477,479,484],{"type":23,"value":478},"The number Anthropic reported from internal use: the share of pull requests receiving substantive review comments rose ",{"type":18,"tag":37,"props":480,"children":481},{},[482],{"type":23,"value":483},"from 16% to 54%",{"type":23,"value":485}," after deployment. The usual caveat applies — that's Anthropic measuring Anthropic, on its own codebase — but it's a concrete, production-shaped result from the same \"several specialized agents in parallel, then synthesize\" pattern that agent teams gives you interactively.",{"type":18,"tag":50,"props":487,"children":489},{"id":488},"getting-started",[490],{"type":23,"value":491},"Getting started",{"type":18,"tag":19,"props":493,"children":494},{},[495,497,503],{"type":23,"value":496},"Agent teams is off by default. Enable it via environment variable, either in your shell or in ",{"type":18,"tag":257,"props":498,"children":500},{"className":499},[],[501],{"type":23,"value":502},"settings.json",{"type":23,"value":504},":",{"type":18,"tag":506,"props":507,"children":511},"pre",{"className":508,"code":509,"language":510,"meta":7,"style":7},"language-json shiki shiki-themes github-dark","{\n  \"env\": {\n    \"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS\": \"1\"\n  }\n}\n","json",[512],{"type":18,"tag":257,"props":513,"children":514},{"__ignoreMap":7},[515,527,542,562,571],{"type":18,"tag":516,"props":517,"children":520},"span",{"class":518,"line":519},"line",1,[521],{"type":18,"tag":516,"props":522,"children":524},{"style":523},"--shiki-default:#E1E4E8",[525],{"type":23,"value":526},"{\n",{"type":18,"tag":516,"props":528,"children":530},{"class":518,"line":529},2,[531,537],{"type":18,"tag":516,"props":532,"children":534},{"style":533},"--shiki-default:#79B8FF",[535],{"type":23,"value":536},"  \"env\"",{"type":18,"tag":516,"props":538,"children":539},{"style":523},[540],{"type":23,"value":541},": {\n",{"type":18,"tag":516,"props":543,"children":545},{"class":518,"line":544},3,[546,551,556],{"type":18,"tag":516,"props":547,"children":548},{"style":533},[549],{"type":23,"value":550},"    \"CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS\"",{"type":18,"tag":516,"props":552,"children":553},{"style":523},[554],{"type":23,"value":555},": ",{"type":18,"tag":516,"props":557,"children":559},{"style":558},"--shiki-default:#9ECBFF",[560],{"type":23,"value":561},"\"1\"\n",{"type":18,"tag":516,"props":563,"children":565},{"class":518,"line":564},4,[566],{"type":18,"tag":516,"props":567,"children":568},{"style":523},[569],{"type":23,"value":570},"  }\n",{"type":18,"tag":516,"props":572,"children":574},{"class":518,"line":573},5,[575],{"type":18,"tag":516,"props":576,"children":577},{"style":523},[578],{"type":23,"value":579},"}\n",{"type":18,"tag":19,"props":581,"children":582},{},[583],{"type":23,"value":584},"Then just describe the team you want in natural language. The docs' example prompt is a good template:",{"type":18,"tag":506,"props":586,"children":589},{"className":587,"code":588,"language":23,"meta":7,"style":7},"language-text shiki shiki-themes github-dark","I'm designing a CLI tool that helps developers track TODO comments across\ntheir codebase. Spawn three teammates to explore this from different angles:\none on UX, one on technical architecture, one playing devil's advocate.\n",[590],{"type":18,"tag":257,"props":591,"children":592},{"__ignoreMap":7},[593,601,609],{"type":18,"tag":516,"props":594,"children":595},{"class":518,"line":519},[596],{"type":18,"tag":516,"props":597,"children":598},{},[599],{"type":23,"value":600},"I'm designing a CLI tool that helps developers track TODO comments across\n",{"type":18,"tag":516,"props":602,"children":603},{"class":518,"line":529},[604],{"type":18,"tag":516,"props":605,"children":606},{},[607],{"type":23,"value":608},"their codebase. Spawn three teammates to explore this from different angles:\n",{"type":18,"tag":516,"props":610,"children":611},{"class":518,"line":544},[612],{"type":18,"tag":516,"props":613,"children":614},{},[615],{"type":23,"value":616},"one on UX, one on technical architecture, one playing devil's advocate.\n",{"type":18,"tag":19,"props":618,"children":619},{},[620],{"type":23,"value":621},"The lead populates the task list, spawns the teammates, and synthesizes when they finish. A few things worth knowing early:",{"type":18,"tag":334,"props":623,"children":624},{},[625,658,668,678],{"type":18,"tag":85,"props":626,"children":627},{},[628,633,635,640,642,648,650,656],{"type":18,"tag":37,"props":629,"children":630},{},[631],{"type":23,"value":632},"Display modes.",{"type":23,"value":634}," By default everything runs ",{"type":18,"tag":125,"props":636,"children":637},{},[638],{"type":23,"value":639},"in-process",{"type":23,"value":641},": teammates appear in an agent panel below your prompt, and you can arrow-key onto one, press Enter to open its transcript, and message it directly. If you prefer each teammate in its own pane, split-pane mode is available via tmux or iTerm2 (",{"type":18,"tag":257,"props":643,"children":645},{"className":644},[],[646],{"type":23,"value":647},"\"teammateMode\": \"auto\"",{"type":23,"value":649}," in ",{"type":18,"tag":257,"props":651,"children":653},{"className":652},[],[654],{"type":23,"value":655},"~\u002F.claude\u002Fsettings.json",{"type":23,"value":657},").",{"type":18,"tag":85,"props":659,"children":660},{},[661,666],{"type":18,"tag":37,"props":662,"children":663},{},[664],{"type":23,"value":665},"You can talk to any teammate directly.",{"type":23,"value":667}," Each is a full session — redirect its approach, ask follow-ups, or stop it, without going through the lead.",{"type":18,"tag":85,"props":669,"children":670},{},[671,676],{"type":18,"tag":37,"props":672,"children":673},{},[674],{"type":23,"value":675},"Plan approval.",{"type":23,"value":677}," For riskier work, ask for teammates to require plan approval: the teammate works read-only until the lead approves its plan, and you can give the lead approval criteria (\"only approve plans that include test coverage\").",{"type":18,"tag":85,"props":679,"children":680},{},[681,686,688,695,697,703,705,711],{"type":18,"tag":37,"props":682,"children":683},{},[684],{"type":23,"value":685},"Quality gates.",{"type":23,"value":687}," ",{"type":18,"tag":26,"props":689,"children":692},{"href":690,"rel":691},"https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fhooks",[30],[693],{"type":23,"value":694},"Hooks",{"type":23,"value":696}," like ",{"type":18,"tag":257,"props":698,"children":700},{"className":699},[],[701],{"type":23,"value":702},"TeammateIdle",{"type":23,"value":704}," and ",{"type":18,"tag":257,"props":706,"children":708},{"className":707},[],[709],{"type":23,"value":710},"TaskCompleted",{"type":23,"value":712}," can block a teammate from going idle or a task from closing until your checks pass.",{"type":18,"tag":19,"props":714,"children":715},{},[716],{"type":23,"value":717},"The docs' sizing advice is sensible: start with 3–5 teammates and roughly 5–6 tasks each, give each teammate ownership of distinct files, and start with read-only work — review and research — before trying parallel implementation.",{"type":18,"tag":50,"props":719,"children":721},{"id":720},"honest-caveats",[722],{"type":23,"value":723},"Honest caveats",{"type":18,"tag":19,"props":725,"children":726},{},[727],{"type":23,"value":728},"This is a research preview, and it behaves like one:",{"type":18,"tag":334,"props":730,"children":731},{},[732,742,759,769,779],{"type":18,"tag":85,"props":733,"children":734},{},[735,740],{"type":18,"tag":37,"props":736,"children":737},{},[738],{"type":23,"value":739},"Token cost scales linearly.",{"type":23,"value":741}," Every teammate is a separate Claude instance with its own context window. A five-agent team burns tokens at something like five sessions' rate. For research and review the trade is often worth it; for routine tasks it isn't.",{"type":18,"tag":85,"props":743,"children":744},{},[745,750,751,757],{"type":18,"tag":37,"props":746,"children":747},{},[748],{"type":23,"value":749},"No session resumption for in-process teammates.",{"type":23,"value":687},{"type":18,"tag":257,"props":752,"children":754},{"className":753},[],[755],{"type":23,"value":756},"\u002Fresume",{"type":23,"value":758}," won't restore them; the lead may try to message teammates that no longer exist.",{"type":18,"tag":85,"props":760,"children":761},{},[762,767],{"type":18,"tag":37,"props":763,"children":764},{},[765],{"type":23,"value":766},"Task status can lag.",{"type":23,"value":768}," Teammates sometimes fail to mark tasks complete, which blocks dependents until you nudge.",{"type":18,"tag":85,"props":770,"children":771},{},[772,777],{"type":18,"tag":37,"props":773,"children":774},{},[775],{"type":23,"value":776},"Structural limits.",{"type":23,"value":778}," One team per session, no nested teams (teammates can't spawn teammates), and the lead is fixed for the session's lifetime.",{"type":18,"tag":85,"props":780,"children":781},{},[782,787],{"type":18,"tag":37,"props":783,"children":784},{},[785],{"type":23,"value":786},"It needs supervision.",{"type":23,"value":788}," The docs are candid: letting a team run unattended for long stretches increases the risk of wasted effort. This is a tool you steer, not one you schedule.",{"type":18,"tag":50,"props":790,"children":792},{"id":791},"where-this-fits-in-an-ai-team-stack",[793],{"type":23,"value":794},"Where this fits in an AI-team stack",{"type":18,"tag":19,"props":796,"children":797},{},[798,800,805,807,813,815,820,822,828,830,836],{"type":23,"value":799},"Agent teams is the most accessible way today to ",{"type":18,"tag":125,"props":801,"children":802},{},[803],{"type":23,"value":804},"feel",{"type":23,"value":806}," what multi-agent collaboration is like: real parallelism, real peer-to-peer coordination, with you in the loop. If that working style clicks, the natural next questions are what a longer-lived ",{"type":18,"tag":26,"props":808,"children":810},{"href":809},"\u002Fblog\u002Fwhat-is-an-ai-employee",[811],{"type":23,"value":812},"AI employee",{"type":23,"value":814}," looks like, and how to go from an interactive team in your terminal to a durable one — which is where ",{"type":18,"tag":26,"props":816,"children":817},{"href":134},[818],{"type":23,"value":819},"production multi-agent frameworks",{"type":23,"value":821}," and shared tool layers like ",{"type":18,"tag":26,"props":823,"children":825},{"href":824},"\u002Fblog\u002Fmcp-explained",[826],{"type":23,"value":827},"MCP",{"type":23,"value":829}," come in. For the bigger picture of assembling agents into a functioning department, start with our guide to ",{"type":18,"tag":26,"props":831,"children":833},{"href":832},"\u002Fblog\u002Fhow-to-build-your-ai-team-2026",[834],{"type":23,"value":835},"building your AI team in 2026",{"type":23,"value":139},{"type":18,"tag":838,"props":839,"children":840},"hr",{},[],{"type":18,"tag":19,"props":842,"children":843},{},[844,846,852],{"type":23,"value":845},"We write about this stuff as it ships. If you're building toward an AI team of your own, follow BuildYour.Group or ",{"type":18,"tag":26,"props":847,"children":849},{"href":848},"\u002F",[850],{"type":23,"value":851},"join the early-access waitlist",{"type":23,"value":853}," — we'd like to build alongside you.",{"type":18,"tag":855,"props":856,"children":857},"style",{},[858],{"type":23,"value":859},"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":7,"searchDepth":544,"depth":544,"links":861},[862,863,868,869,870,871,872,873],{"id":52,"depth":529,"text":55},{"id":142,"depth":529,"text":145,"children":864},[865,866,867],{"id":247,"depth":544,"text":250},{"id":267,"depth":544,"text":270},{"id":290,"depth":544,"text":293},{"id":315,"depth":529,"text":318},{"id":377,"depth":529,"text":380},{"id":449,"depth":529,"text":452},{"id":488,"depth":529,"text":491},{"id":720,"depth":529,"text":723},{"id":791,"depth":529,"text":794},"markdown","content:en:blog:claude-code-agent-teams-guide.md","content","en\u002Fblog\u002Fclaude-code-agent-teams-guide.md","en\u002Fblog\u002Fclaude-code-agent-teams-guide","md",1784513367892]