Deployment Overview
Overview is the Deploy tab’s opening page: eight tiles saying where ContextWorks runs, and the traffic that has actually arrived. You register nothing and maintain nothing here. The page asks the org its questions each time it loads.
The tiles
| Tile | What it checks | Ready means |
|---|---|---|
| MCP | Nothing — the runtime tools ship in the core package | Always ready after install |
| Agentforce | The extension package’s functions exist in the org’s GenAI function registry | The extension package is installed |
| Custom Code | Unpackaged Apex classes exposing @InvocableMethod | Someone has built custom tools on the Entity API |
| Custom Enrichments | Unpackaged classes implementing the provider SPI | Someone has plugged code into payloads |
| Domains | The domain roster | One or more configured; zero warns |
| End Users | Active users holding ContextWorks User | More than zero; zero warns |
| Domain Managers | Active users holding ContextWorks Domain Manager | More than zero; zero warns |
| Status | The package license — Active, Trial, or Expired, with its expiration date | Active; Trial, Expired, or an expiration inside 30 days warns |
A Development status means the org has no package license row — a scratch or source-deployed org. In a subscriber org the tile always shows the real license.
Observed traffic
The bottom of the page shows what has actually called: loads by surface over the last 30 days, as a ranked chart beside a table — surface, loads, tokens, errors, last activity.
Traffic is the only proof that an MCP client is reaching your org. A claude.ai connector or a Claude Code registration lives in the client, not in Salesforce, so the org holds no record that one exists.
The same reading catches custom code that’s written but wired to nothing: the Custom Code tile says the class exists; the traffic table says whether it runs.
What’s next
- Agentforce · MCP · Custom tools
- Usage — the same traffic, in full detail