Does CloudMonitor see my AI prompts or responses?

Q: Does CloudMonitor see my AI prompts or responses?

No. CloudMonitor uses provider Admin and Metrics APIs that return token counts, model names, and identifiers only — never prompt or completion content.

No. CloudMonitor uses each provider’s Admin or Metrics API, which returns token counts, model names, timestamps, and identifiers — never prompt text or completion content.

The specific fields CloudMonitor ingests per provider:

GitHub Copilot — active and engaged user counts, completion and acceptance counts per editor, language, and model, IDE chat counts, pull request summary counts. No source code, no prompts, no chat transcripts.
OpenAI — input tokens, output tokens, cached input tokens, audio tokens, request counts, model name, project ID, user ID, API key ID, and a batch flag. No request bodies.
Anthropic Claude — uncached input, cache write, cache read, and output tokens; service tier; model name; workspace ID; API key ID. No request bodies.
Cursor — paid seat count, per-developer model usage counts, included vs overage split. No editor content.

Provider API keys are encrypted at rest in a tenant-scoped vault and transmitted only over TLS 1.2 or higher. CloudMonitor is ISO 27001, ISO 9001, and ISO 42001 certified.