LevelUpWith

Explore / Building on the Claude Platform: API to Production / Lesson 2

Lesson 2: The Console, Organizations and Workspaces

How an organization is divided into workspaces in the Claude Console, and why the workspace is the unit for isolating environments, resources, spend and rate limits.

Free to read, no account needed. Part of Building on the Claude Platform: API to Production.

By the end of this level you'll be able to

  • Navigate the Console to find the playground, workspace settings, member roles and the rate-limit and spend-limit pages
  • Design a workspace layout (for example dev, staging, production) and explain what the Default Workspace can and cannot do
  • Explain which resources are scoped to a workspace (files, message batches, skills) and why that matters for multi-tenant isolation

Why the workspace matters before you write any code

Every request you send to Claude runs inside exactly one workspace, and that workspace decides three things: which stored resources the request can see, whose budget it draws on, and which rate limit it counts against. If you get the layout wrong, a runaway dev script can eat the throughput your production traffic needs.

The hierarchy is short:

  • Organization: the billing entity. It holds the usage tier, the organization-wide rate limits and the monthly spend cap.
  • Workspace: a partition inside the organization. It has its own members, keys, stored resources and optional lower limits.

If you know AWS, think of the organization as the payer account and workspaces as lightweight member accounts: one bill, separate blast radii.

A tour of the Console

Sign in at platform.claude.com. These are the pages you will use most:

Playground              -> try models and API parameters in the browser
                           (older articles call it the Workbench)
Usage                   -> token, request and rate-limit charts
Rate limits             -> your tier and per-model limits;
                           a Workspace dropdown for per-workspace limits
Settings > Workspaces   -> create, edit, archive; per-workspace spend limits
Settings > Members      -> invite people, add them to workspaces, set roles
Settings > Billing      -> the organization's own spend limit

The workspace picker sits at the top of the sidebar. Leave Settings before you look for it, and check it before you do anything else. Half of all "where did my file go?" confusion is being in the wrong workspace.

Gotcha: rate limits and spend limits live on different pages. Rate limits are on the Rate limits page (pick the workspace from the dropdown, then Edit next to a model). Spend limits are under Settings > Workspaces (select the workspace, then Spend limits in the panel).

The Default Workspace: what it can and cannot do

Every organization starts with a Default Workspace. It works for a first experiment, but it has hard edges:

  • It cannot be renamed, archived or deleted.
  • You cannot set rate or spend limits on it. It always runs at the organization's limits.
  • It has no member list of its own. Access follows each person's organization role, so anyone whose role allows API access can use it.
  • Usage and cost reports show its workspace_id as null, which is easy to misread as "unattributed".

Treat the Default Workspace as a scratchpad. Anything with real traffic or real data belongs in a workspace you created.

You may also see a Claude Code workspace. Anthropic creates it automatically the first time someone signs in to Claude Code with their Console account, to keep that traffic separate. Don't archive it: that disables Claude Code sign-in through Console billing for the whole organization.

Worked example: dev, staging, production

Only organization admins can create workspaces: Settings > Workspaces > Create workspace, then a name and a colour. Here is a layout as a plan (this is a design document, not a file format the platform reads):

organization: acme-support-bot
# Say the org limit for your main model is 2,000,000 input tokens/min.

workspaces:
  dev:
    members: all engineers (Workspace Developer)
    rate_limit: 200,000 input tokens/min   # a looping script can't starve prod
    spend_limit: $200/month                # caps the cost of an experiment
  staging:
    members: engineers (Workspace Developer)
    rate_limit: production-like            # load tests should behave like prod
    spend_limit: $500/month
  production:
    members: two on-call leads (Workspace Admin)
    rate_limit: not set                    # inherits the organization limit
    spend_limit: sized to forecast, with alerts

Three rules make this work:

  1. A workspace limit can be lower than the organization's, never higher.
  2. If you set nothing, the workspace matches the organization limit.
  3. Organization limits always apply, even if workspace limits add up to more. Giving three workspaces 1,000,000 each against a 2,000,000 organization limit does not create 3,000,000 of capacity.

Capping dev also protects the other workspaces, because what dev can't use stays available to production.

Trade-off: archiving a workspace cannot be undone and archives every API key created for it. Name workspaces carefully ("Prod - Support Bot", not "test2") so you never need to. An organization gets 100 workspaces by default; archived ones don't count.

Roles in one table

People have an organization role and, separately, a role in each workspace.

Workspace role What it allows
Workspace User Playground only
Workspace Limited Developer Manage API keys and use the API; no session traces or file downloads
Workspace Developer Manage API keys and use the API
Workspace Admin Full control of workspace settings and members
Workspace Billing View billing; inherited, never assigned by hand

Inheritance is where people trip up. Organization admins are Workspace Admin everywhere, and organization billing members get Workspace Billing everywhere. Organization users and developers get nothing automatically: an admin must add them on the Members page with Add to workspace. That is how you keep most engineers out of production.

What is scoped to a workspace, and why tenants care

Three kinds of stored resource belong to the workspace they were created in:

  • Files uploaded through the Files API
  • Message Batches
  • Skills

Prompt caches are also isolated per workspace on the Claude API. Workspaces themselves and organization members are managed at the organization level.

The consequence: a file is visible to the whole workspace, not to one end user or conversation. Any key with access to the workspace can reference any file in it. If customer A's and customer B's documents share a workspace, only your application code keeps them apart.

You have two defences:

  • Always: treat file IDs as server-side references. Never accept one from an end user. Keep the user-to-file mapping in your own database.
  • For hard isolation: give each tenant their own workspace. The boundary is then enforced by the platform rather than by your code. The 100-workspace default means this suits tens of large tenants, not thousands of small ones, unless your account team raises the limit.

Key takeaways

  • Every request runs in exactly one workspace, which sets resource visibility, spend and rate limits.
  • The Default Workspace cannot be renamed, archived, deleted or given limits. Use it as a scratchpad only.
  • Workspace limits can only go below organization limits, and organization limits always win.
  • Rate limits are set on the Rate limits page; spend limits under Settings > Workspaces.
  • Organization admins and billing members inherit access everywhere; users and developers must be added per workspace.
  • Files, Message Batches and Skills are workspace-scoped, so the workspace is your isolation boundary for environments and tenants.

A 6-question knowledge check follows this lesson in the app, free and without an account.

Keep going: levels 3 and 4 are free

Create a free account to open the next two levels, save your progress and pick up where you left off. No card needed.

Next: Level 3, API Keys and Authentication.

Create a free account →See the whole course

More like this

  • MCP in Depth: Build, Secure and Run Servers and Clients

    Build Model Context Protocol servers and clients in Python, from a first tool to an authorised remote deployment. You will learn the stateless 2026-07-28 revis…

    Course · Claude · AI & Agents · Expert · 47 levels · 4 free · Professionally curated by levelupwith.com

  • Production Agents on Claude: Design, Build, Run

    Design, build and operate Claude agents that hold up in production: choose between workflows and agents, use the Claude Agent SDK and Managed Agents well, and…

    Course · Claude · AI & Agents · Expert · 64 levels · 4 free · Professionally curated by levelupwith.com

  • Claude at Work: No-Code Skills for Knowledge Workers

    Learn to brief Claude clearly, check what it gives you, and set up Projects, Artifacts, Skills, memory, connectors and Research for real work in sales, marketi…

    Course · Claude · AI & Agents · Capable · 18 levels · 4 free · Professionally curated by levelupwith.com

  • Delegating Real Work to Claude: Tasks, Files and Apps

    Learn to hand Claude multi-step work on your real files, browser and Office apps, with clear briefs, firm limits and approval gates, then review the results so…

    Course · Claude · AI & Agents · Capable · 18 levels · 4 free · Professionally curated by levelupwith.com

  • Prompting Claude: Instructions, Context and Testing

    Write prompts that Claude follows reliably, manage what goes into its context window, and test and improve your prompts with evidence instead of guesswork. Eac…

    Course · Claude · AI & Agents · Capable · 20 levels · 4 free · Professionally curated by levelupwith.com

  • Finished Work with Claude Artifacts, No Code

    Brief, refine, brand, check, share and export finished designs, decks, documents, dashboards and motion pieces in Claude without writing code. Every level ends…

    Course · Claude · AI & Agents · Capable · 18 levels · 4 free · Professionally curated by levelupwith.com