> ## Documentation Index
> Fetch the complete documentation index at: https://docs.costhawk.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Admin API Sync

> Connect Anthropic/OpenAI admin APIs for low-friction org-level cost visibility

Admin API sync is the fastest path to organization-wide cost visibility.

It gives you aggregate usage and cost data without changing your application code.

<Card title="MCP Still Supported" icon="terminal" href="/mcp-server/overview">
  MCP is not being removed. Keep using MCP tools for local workflows while Admin API sync handles org-level provider reporting.
</Card>

## When To Use This

* You want quick visibility with minimal setup
* You manage usage for a team or organization
* You do not need per-request hard-stop enforcement yet

## What You Get

* Anthropic usage buckets (`/v1/organizations/usage_report/messages`)
* Anthropic cost buckets (`/v1/organizations/cost_report`)
* Anthropic Claude Code analytics (`/v1/organizations/usage_report/claude_code`)
* OpenAI usage buckets (`/v1/organization/usage/completions`)
* OpenAI daily costs (`/v1/organization/costs`)

## Setup

<Steps>
  <Step title="Create provider admin keys">
    Create an Anthropic Admin key (`sk-ant-admin-...`) and an OpenAI organization Admin key in each provider console.
  </Step>

  <Step title="Connect keys in CostHawk">
    In CostHawk, add each connection under Integrations/Connections and run a connection test.
  </Step>

  <Step title="Wait for first sync">
    Initial sync usually appears within 5 minutes. Cost endpoints are daily-granularity by provider design.
  </Step>
</Steps>

## Sync Cadence

* Usage endpoints: every 5 minutes
* Cost endpoints: every 5 minutes (daily buckets)
* Claude Code analytics: hourly

## Security Model

* Admin keys are encrypted at rest
* Plaintext keys are never returned after creation
* Connection list only shows fingerprint + provider org metadata
* Connection lifecycle actions are audit logged

## What Admin API Sync Does Not Replace

* **Hard spending limits / request blocking**: use proxy + wrapped keys
* **Google Gemini real-time coverage**: use proxy (Google has no equivalent usage API)

<CardGroup cols={2}>
  <Card title="Integration Overview" icon="layers" href="/integrations/overview">
    Compare Admin API, OTel, and Proxy paths
  </Card>

  <Card title="Google Gemini Proxy" icon="google" href="/integrations/google">
    Required path for Google real-time tracking
  </Card>
</CardGroup>