Connect
Create an API token in the Pagerless app, then add the server to your client. It is stateless Streamable HTTP with a bearer token.
Claude Code
claude mcp add --transport http pagerless https://app.pagerless.dev/mcp --header "Authorization: Bearer plt_your_token"Any MCP client (JSON config)
{
"mcpServers": {
"pagerless": {
"url": "https://app.pagerless.dev/mcp",
"headers": { "Authorization": "Bearer plt_your_token" }
}
}
}Tools
| Tool | What it does |
|---|---|
problem_list | Lists problems, open ones by default. |
problem_get | One problem in full: the diagnosis, up to 5 scrubbed samples and recent deploys. App text is marked untrusted. |
action_propose | Proposes a fix. The answer is runs alone, waiting for a human, or refused, decided by your rules. |
action_status | Where a proposed fix stands. |
action_report | After the agent carried out an approved fix itself (say, a code change), reports whether it worked. |
human_ask | Asks the on-call person one yes/no question. A yes never bypasses policy. |
human_answer_get | Reads the answer to a human_ask. |
verify | Watches the problem's error rate for 10 to 120 seconds (default 60). |
problem_resolve | Closes a problem with a plain-words summary. |
Limits and safety
- 120 requests per minute per token.
- Every call is written to the audit log as the agent's token.
- Agents never act on production directly: a fix runs only if your rules allow it, or a person approves it.
- Discovery file for agents: /.well-known/mcp.json.
Questions
Which agents work with it?
Any client that speaks MCP over Streamable HTTP with a bearer header, such as Claude Code, Claude Desktop, Cursor and VS Code.
Can the agent roll back production on its own?
Only if your rules set that fix kind to alone. When a person has to approve an undo, the approval always needs a passkey.