Pingera MCP Server
OfficialEnables configuring and testing Pingera alert channels that send notifications to Telegram, so monitoring checks can alert users in Telegram after failures.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
Pingera MCP Server
A Model Context Protocol server for Pingera: let AI assistants investigate and manage uptime, API, SSL, DNS, port-scan and Playwright checks, alerts, heartbeat monitors and status pages.
It runs two ways:
Self-hosted - on your machine over stdio (Claude Desktop, Cursor, VS Code, ...) or as your own HTTP server, with your Pingera API key.
Multi-tenant - the hosted server at
https://mcp.app.pingera.ru/mcp(or your own deployment): one HTTP server for many users, where every request carries the user's own Pingera credentials.
Built on the official MCP Python SDK 2.x and the Pingera Python SDK.
Quick start
You need a Pingera API key: create one at app.pingera.ru/settings/tokens.
Hosted server (nothing to install)
Connect your MCP client to https://mcp.app.pingera.ru/mcp (streamable HTTP) and send your key in the
Authorization header. Claude Code:
claude mcp add --transport http pingera https://mcp.app.pingera.ru/mcp --header "Authorization: Bearer YOUR_PINGERA_API_KEY"Cursor (.cursor/mcp.json; VS Code's .vscode/mcp.json is the same under "servers" with "type": "http"):
{
"mcpServers": {
"pingera": {
"url": "https://mcp.app.pingera.ru/mcp",
"headers": {
"Authorization": "Bearer YOUR_PINGERA_API_KEY",
"X-Pingera-Toolsets": "default,heartbeats"
}
}
}
}The optional X-Pingera-Toolsets, X-Pingera-Tools and X-Pingera-Mode: read_only headers choose which tools this
client gets (see Tools and Per-request options).
Local server over stdio
Requires uv. Claude Desktop (claude_desktop_config.json), Cursor (.cursor/mcp.json)
and similar clients:
{
"mcpServers": {
"pingera": {
"command": "uvx",
"args": ["--from", "pingera-mcp-server", "pingera-mcp"],
"env": {
"PINGERA_API_KEY": "YOUR_PINGERA_API_KEY",
"PINGERA_MODE": "read_write"
}
}
}
}Leave out PINGERA_MODE to start read-only. From a checkout: uv sync && uv run pingera-mcp.
Then ask things like "Which checks failed in the last 24 hours and why?", "Create an API check for https://api.example.com/health from Moscow and Frankfurt and alert me in Telegram after 3 failures" or "Post an incident on our status page: payments are degraded".
Related MCP server: OpenAPI MCP Server
Tools
Tools are grouped into toolsets. The default toolsets cover monitoring, alerting and status pages (83 tools in read-write mode, 50 read-only; 126 with all toolsets). Fewer tools means a smaller prompt and better tool choice, so enable only what you use:
Where | Choose toolsets | Choose exact tools |
Hosted or any HTTP server, per client |
|
|
Your own server (stdio or HTTP default) |
|
|
Toolset names are listed below; default and all work everywhere. A header overrides the server's default for that
client; exact tools override toolsets. Selection is per connection (headers or server settings), not per tool call:
clients load the tool list once, before the model uses any tool.
Toolset | What | Tools |
| Connection check and reference documentation (always enabled) |
|
| Organization details and usage |
|
| Status pages |
|
| Status page components, component groups and component uptime |
|
| Status page incidents, incident updates, scheduled maintenance, postmortems and incident templates |
|
| Monitoring checks: configuration, regions and on-demand runs |
|
| Check results, uptime, response times, status history and multi-region executions |
|
| Alerts, alert rules and alert channels |
|
| Heartbeat (cron job) monitors and their pings |
|
| Check groups |
|
| Organization secrets and the secrets injected into checks |
|
| Status page passwords and custom sender email domains |
|
| Private locations and runner tokens for self-hosted check runners |
|
| PDF uptime reports |
|
| Organization audit log |
|
Access levels. Read tools only read. Runs execute something without changing configuration (run a check now,
send a test notification, request a report); they are available in read-only mode. Changes create, update or delete
data and are hidden in read-only mode. Tool annotations (readOnlyHint, destructiveHint, idempotentHint) tell
clients which tools need confirmation.
Arguments. Create and update tools are generated from the Pingera SDK models, so their schemas carry the API's
field descriptions, limits and allowed values. Update tools change only the fields you pass. Time arguments accept ISO
8601 or relative values like 30m, 24h, 7d.
Also available: resources pingera://reference/check_types and pingera://reference/alerts (also through the
get_reference tool) and prompts investigate_check, create_synthetic_check and communicate_incident.
Configuration
Environment variables (a .env file is read too; see .env.example). Command line options
(pingera-mcp --help) override them.
Variable | Default | Meaning |
| - | Credentials for stdio and for HTTP with |
|
| Pingera API (a trailing |
|
|
|
|
| Comma-separated toolsets, |
| - | Comma-separated tool names to expose instead of toolsets, e.g. |
|
| API request timeout (seconds) and retries |
|
|
|
|
| HTTP listener and endpoint |
|
| HTTP credentials: |
|
| Answer HTTP requests with JSON instead of SSE streams |
| - | Only accept these Host / Origin values (DNS rebinding protection; |
|
| Logging level ( |
HTTP server
pingera-mcp --transport http --mode read_write # development, http://127.0.0.1:8000/mcp
uvicorn pingera_mcp.asgi:app --host 0.0.0.0 --workers 4 # production (configured by environment)
docker compose up # the Docker image, see DOCKER.mdThe endpoint is /mcp; /health is the health check and / describes the server. The server is stateless: it
keeps no MCP sessions, so any worker or replica can answer any request and it scales horizontally without sticky
sessions.
Credentials
With the default MCP_HTTP_AUTH=request (multi-tenant), every request must carry Pingera credentials, otherwise the
server answers 401:
Header | Meaning |
| API key (works with clients that only support bearer tokens) |
| Pingera JWT access token (recognized by its format) |
| API key, the way the Pingera API itself accepts it |
Credentials are used only for that request's API calls; they are never stored or shared between requests.
With MCP_HTTP_AUTH=env the server uses its own PINGERA_API_KEY for every client. Use it only for a private,
single-account deployment.
Per-request options
Header | Effect |
| Hide and block tools that change data for this client (cannot lift a read-only server) |
| Use these toolsets instead of the server default ( |
| Expose exactly these tools |
Development
uv sync --extra dev
uv run pytest --ignore=tests/e2e # unit and integration tests (fake Pingera API, no network)
PINGERA_E2E_URL=http://localhost:8000/mcp PINGERA_E2E_API_KEY=... uv run pytest tests/e2eReleasing
Bump __version__ in pingera_mcp/__init__.py and add a ## <version> section to CHANGELOG.md, merge to main,
then push a matching tag:
git tag 0.2.0 && git push origin 0.2.0The release workflow runs the tests, builds and smoke-tests the package, publishes it to PyPI (Trusted Publishing) and creates a GitHub release with the changelog section as notes. The Docker image is built from the same tag with build/cluster-build.
The end-to-end tests run against a real server and API and create and delete objects in that organization, so use a test organization. docs/architecture.md explains how the server is put together and how to add tools. scripts/smoke_client.py calls a tool on a running server, and examples/langgraph_agent.py connects a LangGraph agent to it.
License
MIT - see LICENSE.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
The Mercado Pago MCP Server implements the Model Context Protocol to provide AI agents and LLMs with access to Mercado Pago's APIs and tools within compatible development environments. It acts as an intermediary that translates Mercado Pago resources into executable functions (tools) that AI applications can invoke to perform actions and automate flows. The server simplifies integration, enables using documentation to implement or improve code, and optimizes operations through natural language interactions without manual implementations.
AI-powered incident management and server monitoring via MCP.
MCP server for building and testing AI agents with multi-model experimentation and insights.
A Model Context Protocol server for Wix AI tools
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceA server that implements the Model Context Protocol, providing a standardized way to connect AI models to different data sources and tools.12MIT
- AlicenseNot gradedqualityDmaintenanceA server that implements the Model Context Protocol (MCP) with StreamableHTTP transport, enabling standardized interaction with model services through a RESTful API interface.2MIT
- AlicenseBqualityDmaintenanceA Model Context Protocol server that enables AI assistants to query Prometheus metrics, discover available data, and analyze system performance through natural language interactions.5236 npmMIT
- AlicenseNot gradedqualityDmaintenanceA Model Context Protocol (MCP) server for comprehensive monitoring and observability of systems using Langfuse.1MIT