Announced Nov 10, agents can pick hardware, deploy Pods or Serverless endpoints, diagnose failures and shut resources down, with OAuth sign-in and no API key on supporting clients. Every MCP tool is marked read-only or destructive, so compatible clients can ask for confirmation before deletions.
Announced Nov 10, it lets teams ask compatible AI assistants to compare studies, pull participant quotes and read existing AI Insights and reports. Answers link back to the original session recordings…
Announced Nov 10, it exposes Cast AI costs, recommendations and cluster policies as tools any MCP compatible client can call. Access is read only, so an agent can investigate production problems but not change clusters…
Announced Nov 10, it's listed in the ChatGPT Plugins and Microsoft Copilot Connectors directories, and Claude and other MCP clients connect through tdmcp.com. You can run a test send before a real one, and opt outs are honored automatically…
Announced Nov 10, it installs npm and Python packages, CLIs and remote URLs as MCP servers through one interface, and checks that each works before an agent uses it. API tokens go in the system keychain instead of config files; it's MIT licensed and runs on macOS and Linux…
Launched Nov 10, you paste one line into your agent, sign in with OAuth (no API key), and it deploys, reads logs and rolls back. Billing is per minute from prepaid wallet credit ($5/mo Personal, $20/mo Team), with a monthly cap that pauses apps…
Spotted something we missed? Start a thread.
