opsgenie-mcp

opsgenie-mcp

MCP server for Opsgenie (Atlassian's incident/alert management andon-call platform) — exposes the full public Opsgenie REST API as MCP tools.

⚠️ Vendor note: Atlassian has announced Opsgenie is reachingend of supportand recommends migrating to Jira Service Management or Compass. The publicAPI documented here (docs.opsgenie.com) is still live and unchanged as ofthis build, but expect it to eventually be deprecated.

MSPbots Connector gives your AI clients one MCP URL to reach your entire MSP stack (RMM, PSA, documentation, security, finance), and gives you control over who can call which tool and a record of everything that was called.

Hosted service: connect.mspbots.ai · Free to use.

Why Connector

Quick start

1. Sign up at connect.mspbots.ai.
2. Connect your vendors. Pick from 150+ connectors and authorize each one on the vendor's own page, or bring your own API keys. No extra vendor accounts needed.
3. Set access. Create roles and decide which tools each role may call. Nothing is allowed until you allow it.
4. Add your MCP URL to your AI client. Copy the URL from your Connector dashboard into Claude, ChatGPT, Cursor, VS Code, Gemini, Codex, or any MCP-capable client.
5. Ask your AI to do real work, and watch every call appear in the audit log.

Features

Unified gateway

• One MCP URL for all your AI clients
• 150+ connectors across the MSP stack
• Custom MCP servers supported, free
• Authorization happens on each vendor's own page; the AI never sees passwords or tokens
• Bring your own credentials: use the vendor API keys you already have
• Multiple accounts per app; connect once, reuse across AI clients
• Tool search: finds the right action for the task
• One sign-in; vendor credentials are injected behind the scenes and OAuth tokens refresh automatically

Visibility

• Audit log for every tool call (who, which vendor, when)
• Usage dashboard: call volume, top vendors, error rate, per-user activity
• Export audit logs to CSV, or stream to Loki, Graylog, or LogScale (SIEM)
• Optional recording of tool arguments and result summaries (off by default, enabled per organization)

Control

• Per-role, per-tool allowlist, deny by default
• owner / admin / member roles
• SCIM provisioning with Entra ID, Okta, JumpCloud, and Google Workspace

Security

• Credentials encrypted one record at a time (AES-256-GCM, per-record key)
• AI clients authenticate with OAuth 2.1 + PKCE
• Vendor responses are scanned for prompt-injection patterns and returned as structured data. This is one layer of defense, not a guarantee against every attack.
• Static egress IP for customer firewall and API allowlisting

Pricing

Free.

Links: connect.mspbots.ai · mspbots.ai

Overview

• Stateless HTTP service. No credentials are ever persisted — each requestsupplies its own API key via a header, used only for the lifetime of thatsingle request.
• Supports concurrent requests; per-request credential isolation is done viaPython contextvars, not a global/shared client instance.
• Entry points: POST /mcp (MCP protocol) and GET /health (health check).
• Default port: 8080 (configurable via MCP_HTTP_PORT).

Scope

22 tools, trimmed down from an original 213-tool full-API build(2026-08-04), all in the single alert category. MSPbots' own storedintegration config for this vendor calls exactly 1 endpoint(GET /v2/alerts, read-only, → opsgenie_alert_list_alerts). Per the"actual usage + same-category core CRUD" scope decision, every othercategory (Incident, Incident Timeline, Integration, Heartbeat, Alert &Notification Policy, Policy (legacy v1), Maintenance, Account, User,Custom User Role, Contact, Notification Rule (+ Step), Team (+ Member/Role/Routing Rule), Schedule (+ Rotation/Override), Escalation, Who Is OnCall, Forwarding Rule, Service (+ Incident Rules/Templates), IncidentTemplates — 27 categories, ~178 tools) was removed entirely, sinceMSPbots doesn't touch any of them; within the alert category itself,35 original tools were cut to the 22 core alert-lifecycle operations(list/get/create/close/delete/acknowledge/unacknowledge/note/tags/assign/responder/team/escalate/snooze/update message-description-priority/count/logs/request-status) — dropped were the saved-search sub-feature (5tools), the attachment sub-feature (4 tools), custom-properties add/remove(2 tools), execute_custom_action (1), and list_alert_recipients (1),none of which are core alert CRUD.

Source data for the kept tools was originally extracted by fetchingdocs.opsgenie.com/docs/alert-api and structuring each documentedoperation (method, path, path/query params, JSON body fields) into amachine-readable format — the same codegen-from-structured-spec approachused for other large-API vendors in this program (ConnectSecure, Dynu,Jira Data Center), adapted here because Opsgenie has no downloadableOpenAPI/Postman spec, only prose documentation. If a removed category isneeded later, the same source pages (docs.opsgenie.com/docs/*-api) canbe re-parsed the same way.

Authentication

Opsgenie uses a static API key tied to a specific Opsgenie integration(created in Opsgenie's Settings → Integrations → API). MSPbots' ownintegration convention sends this key as Authorization: GenieKey <apiKey>,matching Opsgenie's own documented format, and this server forwards itexactly that way.

HEADER authentication parameters

Missing the required header returns 401:

{
  "error": "Missing credentials",
  "message": "This server requires the X-Opsgenie-Api-Key header",
  "required_headers": ["X-Opsgenie-Api-Key"],
  "optional_headers": ["X-Opsgenie-Base-Url"]
}

Environment Variables

MCP Endpoint

• POST /mcp — MCP protocol (streamable HTTP transport)
• GET /health — health check, returns {"status": "ok"} (pure local liveness probe, does not call Opsgenie)

Tool List

Tool names are opsgenie_<category>_<operation>, derived from eachoperation's heading in the official docs (e.g. "List Alerts" in the alertcategory → opsgenie_alert_list_alerts). body parameters are accepted asa generic dict — the exact field list for each is documented in thattool's own docstring (extracted from the source docs), and the full fieldschema is available in Opsgenie's own API reference (linked below).

Test Example

# Health check
curl -s http://localhost:8080/health

# Call a tool via the MCP protocol (streamable HTTP) — requires an
# initialize handshake first per the MCP spec; abbreviated example below
# shows the tool-call request body only:
curl -s -X POST http://localhost:8080/mcp \
  -H "X-Opsgenie-Api-Key: <your-opsgenie-api-key>" \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "mcp-session-id: <session-id-from-initialize>" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "opsgenie_alert_list_alerts",
      "arguments": {"limit": 5}
    }
  }'

Live-verified (2026-07-30): opsgenie_alert_list_alerts (the sameendpoint MSPbots itself calls) was called end-to-end through this runningserver with a real test API key and returned real alert data (message,status, priority, responders, integration, etc.) from the account. Twoother tools in different categories (opsgenie_team_list_teams,opsgenie_account_get_account_info) were also called with the same key andcorrectly returned Opsgenie's own 403 API Key is not granted with configuration access — confirming the request/auth-header plumbing workscorrectly outside the alert category too; this specific test key is justscoped to alert-level access only (a normal Opsgenie integration-keyrestriction, not a bug).

API Reference

• Overview: https://docs.opsgenie.com/docs/api-overview
• Authentication: https://docs.opsgenie.com/docs/authentication
• Per-category docs are linked from the sidebar at https://docs.opsgenie.com/docs/alert-api

Known Gaps

• Trimmed from 213 to 22 tools on 2026-08-04. The original buildcovered the full public API across 28 categories per an earlier scopedecision. A later scope decision cut this back to MSPbots' actually-usedcategory (alert) plus its core CRUD — see the Scope section above forexactly what was kept/dropped within alert and the full list of the 27removed categories (~178 tools). If a removed category is needed later,the same docs.opsgenie.com/docs/*-api pages can be re-parsed the sameway the kept tools were generated.
• Several kept tools still mutate real Opsgenie alert state(opsgenie_alert_delete_alert, opsgenie_alert_close_alert,opsgenie_alert_create_alert, etc.) — treat these asirreversible/state-changing and confirm with a human before invoking.
• body parameters are untyped (dict) rather than fully modeled — eachtool's docstring lists the field names extracted from Opsgenie's docs, butreproducing all of them as typed Python parameters was out of scope for amechanically-generated server.
• Source data was extracted from prose documentation, not a machinespec — Opsgenie has no downloadable OpenAPI/Postman collection, so everyoperation's method/path/params was parsed from docs.opsgenie.com's HTML.
• Only opsgenie_alert_list_alerts (MSPbots' own endpoint) waslive-verified with real data returned. The remaining 21 tools arestructurally correct (schema validated, MCP-protocol tools/listconfirmed, all pass compile()) but not individually smoke-tested —several are write/destructive operations that would create, modify, orclose real alerts, so they weren't exercised against the live testaccount.
• Vendor is reaching end-of-support (see the warning banner at the topof this README) — Atlassian recommends migrating to Jira ServiceManagement or Compass. This MCP targets the still-live Opsgenie public APIas of this build.