At the start of every session
Once CheckMyApp is connected to your agent, it tells the agent, the moment it connects, how your watched apps did in their latest checks and which findings are new. Your agent raises them before you ask, and offers to fix them.
You can also ask at any time:
What did CheckMyApp find since my last session?
The agent answers with latest_results: every app’s last check, its verdict, and what is new since the check before.
Pushed into a running session (preview)
For a Claude Code session you leave open, CheckMyApp pushes each finished Daily Watch check straight into it: the app, the verdict and the findings that are new. The agent tells you and offers to fix them — it does not start changing anything until you say so. Opening a session does not replay old results; a check with new findings from the last 24 hours is the one exception, and it is waiting for you when you sit down.
Add it once, with your API key in place of <KEY>:
claude mcp add checkmyapp-watch -e CHECKMYAPP_API_KEY=<KEY> -- npx -y https://checkmyapp.dev/mcp/checkmyapp-watch-1.0.0.tgzIt builds on Claude Code channels, which are a research preview; while they are, a channel like ours is switched on per session with a flag:
claude --dangerously-load-development-channels server:checkmyapp-watchKeep CheckMyApp itself connected as well: the push announces, and the agent fixes from the full review it asks CheckMyApp for. On claude.ai Team and Enterprise plans, channels must first be enabled by an Owner of your organization.
not using an agent?
Every result also stays in your dashboard. From the app’s settings, new regressions can become tickets in Linear, and results can go to email, Slack or your own webhook.