Skip to main content
Questions that come up often during evaluation, onboarding and building, grouped by topic. If yours isn’t here, check Troubleshooting for symptom-by-symptom fixes, Error Codes for a specific error message, or Concepts & Glossary for terms.

Getting started

Dimension by ThreatAware is an AI-powered workspace for building your own security apps. You connect Data Feeds, build Studio apps by describing what you want, and wire up Actions that send webhooks or emails. See What is Dimension?.
Go to https://<your-tenant>.dimension.io and sign in with the same account you use for the main ThreatAware platform. Your role there decides what you can do in Dimension, and MFA and SSO carry across. See First steps.
No. Every pillar has a Composer: you describe what you need in plain language and it builds the configuration or the app. If you want the code, you can view it in X-Ray and export the full React/TypeScript source. See Exporting.
Import one of the five pre-built apps. Each runs on demo data until you add your own credentials. To build from scratch, follow Build your first app.

AI and models

Inside Dimension, the Planner and Composers use Anthropic’s Claude models. You pick one from the picker next to the prompt box:
  • Opus 5.5 (recommended)
  • Sonnet 5.5
  • Haiku 5.5
  • Fable 5.1
To build with another model, use it from your own AI assistant and connect that assistant to Dimension through MCP. For example, OpenAI Codex or GitHub Copilot CLI can build data feeds, actions and apps in Dimension with whichever model the assistant uses. See Connect with MCP.A Super Admin controls which models people can pick under Settings → AI → Composer models. Hover a model there to see what it costs in AI credits. Two other settings on that page also shape how models are used:
  • Cross-origin models (off by default) allows Claude versions that aren’t available in your region. Requests to those models may be processed outside your region.
  • AI data access (off by default) decides whether the AI sees only your data’s field names or the values as well.
See Settings → AI.
The chat assistant on this docs site is separate from the AI inside Dimension. If you ask it “what model are you?”, it answers about itself, not about Dimension.
Not by default. The Composer sees the schema of your data, such as field names and types, but never the values. A Super Admin can turn on AI data access under Settings → AI if your data-handling policies allow it. See Schema-only AI access.
No. By default we never read, share or train models on your security data. AI providers process Composer prompts under their published data-handling agreements. See Security & Privacy.
They do different jobs and work together. An AI assistant such as Claude Code, Claude Desktop, Copilot or Codex is where you have the conversation. Dimension is where the result lives: your data feeds, Studio apps and actions, hosted with SSO, roles, audit logging and human approval for outbound actions.You can build in Dimension directly with its own Composers, or connect your assistant to Dimension through MCP and build from there. Either way, the assistant can only do what your Dimension role allows. See Connect with MCP.
Yes. The Dimension MCP server works with Claude Code, Claude Desktop, Visual Studio Code, Visual Studio, GitHub Copilot CLI and OpenAI Codex. Open your avatar menu → Connect MCP to copy the setup for your tenant. See Connect with MCP.
Use the Planner when you have a goal but aren’t sure what to build: it breaks the goal into data feed, action and app steps with ready-made prompts. Use a Composer directly when you already know what you need.
AI credits are your allowance for Composer and Planner work. Track them under Settings → Usage. Unused credits don’t roll over to the next billing period. Restoring a checkpoint costs nothing, so it’s cheaper than re-prompting. See Settings → Usage.

Building

Yes, as Studio apps. An app can collect input in a form, then fire Actions that open a ticket, post to Slack or send an email, from a button click, a form submit or conditional logic. Inbound webhook triggers, where outside systems start an action, are on the roadmap.
Any source with an API, such as CrowdStrike, Jira or ServiceNow. The Data Feed Composer handles authentication, pagination, schema mapping and scheduling. See Connect your data.
Only people in the app’s workspace who have access through its Access list, and only once the app is published. Building or sharing an app never gives anyone access to data they couldn’t already see. See Publishing.
Yes. In the Studio editor, use Restore checkpoint in the Composer chat to go back to any earlier turn. See Restore checkpoint.
Not yet. The last save wins, so agree who’s editing before you start.
X-Ray is read-only, so changes go through the Composer. To edit by hand, export the app, change it in your own editor, then import it back.
Yes. Export it or use X-Ray → Download Package to get the full React/TypeScript source, which runs anywhere Node.js and React do. You’ll need to point it at your own data source, because the Dimension data feeds don’t come with it.
Actions belong to one workspace. To reuse one, migrate it, or export it and import it into the other workspace. See the Migration playbook.

Security and access

On AWS, in the regions your ThreatAware tenant is hosted in. Every tenant is isolated, and workspaces can’t read each other’s data. See Architecture.
Apps run in a sandbox with no internet access and can only read the data feeds connected to them. The only way to send data out is a sanctioned Action, which a person approves. A Super Admin can let users skip repeat approvals with Allow approved actions. See Approving actions. See Apps are secure by design.
Super Admins only. The settings apply to every workspace in your tenant. See Settings → AI.
ISO 27001. See Certifications for audit status and how to request evidence.

Plans and access

Settings → Usage shows your plan, AI credits, and quotas for feeds, records, actions, action runs, apps and users. Click Manage plan to raise your limits.
Not yet. A public Dimension API is on the roadmap. Until then, you can work with Dimension programmatically through the MCP server.

Cross-references

Troubleshooting

Symptom-driven fixes

Error Codes

Error codes and messages

Concepts & Glossary

Every term Dimension uses

First steps

Your first 15 minutes

Security & Privacy

Isolation, AI controls, data handling

Support

When the docs don’t have it