FireHydrant MCP Server – Incident Response Tools
FireHydrant's official MCP server lets AI assistants interact with the FireHydrant API for incident response workflows. Agents can list and create incidents, review alerts, and work with retrospectives when authenticated with a FireHydrant API key.
Overview
FireHydrant MCP Server is FireHydrant's official Model Context Protocol server for connecting MCP-compatible AI assistants to the FireHydrant incident management platform. It is implemented as a Node/TypeScript MCP server and is distributed through the public firehydrant/firehydrant-mcp repository and the firehydrant-mcp command-line package.
The server is designed for responders who want to use AI tools while staying close to incident context. Instead of manually switching between an IDE, chat assistant, and FireHydrant, an AI agent can call documented MCP tools that operate against the FireHydrant API using the permissions of the supplied API key.
What the MCP server enables
The current desktop extension manifest exposes FireHydrant tools for listing incidents, creating incidents, listing retrospectives for an incident, updating retrospective fields, and listing alerts. These capabilities support common response and post-incident workflows, such as checking active incidents, opening a new incident from an investigation, reviewing Signals and third-party alerts, and updating RCA or retrospective details.
The server does not make the model omniscient: it only exposes the FireHydrant API operations implemented by the MCP package and authorized by the configured FireHydrant credentials. Any access restrictions, rate limits, and audit expectations should be handled according to the FireHydrant account and API key policies used by the organization.
When to use it
Use FireHydrant MCP when an incident responder or platform engineer wants an AI assistant to help summarize operational context, retrieve open incidents, create an incident from a discovered problem, or draft and update retrospective information. It is especially useful in IDE-based or assistant-based workflows where responders need structured incident data without leaving their current tool.
Connection and authentication
FireHydrant documents a local stdio setup using npx firehydrant-mcp start --api-key .... The same command pattern is shown for Cursor, Claude Code CLI, Windsurf, VS Code, and Claude Desktop. FireHydrant also publishes an MCP Bundle/Desktop Extension that packages the server and configuration for Claude Desktop-style installation. The API key should be stored as a secret in the user's MCP client configuration and rotated according to the organization's security practices.
Key considerations
The documented transport for the MCP server is local stdio through the npm package. FireHydrant's README notes that Claude Desktop does not yet support SSE or remote MCP servers in that setup, so this YAML intentionally includes only the stdio transport. Because the server can create incidents and update retrospective fields, teams should use least-privilege API keys and review generated changes before relying on agent-driven actions during production incidents.
Supported Transports
stdio
Command: npx
Args:
firehydrant-mcpstart--api-key${FIREHYDRANT_API_KEY}
Frequently Asked Questions
- When should an AI agent use the FireHydrant MCP server?
- Use it when an AI assistant needs structured access to FireHydrant incident response data, such as listing incidents, checking alerts, creating a new incident, or helping with retrospective workflows from an MCP-compatible IDE or assistant.
- What does this MCP server add to an AI agent's capabilities?
- It gives the agent live FireHydrant API-backed tools instead of relying on static model knowledge. The documented tools include listing incidents, creating incidents, listing incident retrospectives, updating retrospective fields, and listing alerts.
- What can the AI agent access or manage through this server?
- The official desktop extension manifest lists tools for incidents, alerts, and retrospectives. The agent can only perform operations implemented by the FireHydrant MCP server and allowed by the configured FireHydrant API key.
- How is authentication configured for FireHydrant MCP?
- FireHydrant's documented local configuration passes an API key to the server with the `--api-key` option. Store the value in a client-side secret such as `FIREHYDRANT_API_KEY`; do not commit real keys to configuration files or repositories.
- Which transport should be used for this MCP server?
- Use the documented local stdio transport started with `npx firehydrant-mcp start --api-key ...`. The README specifically notes that Claude Desktop does not yet support SSE or remote MCP servers in its documented setup, so no SSE or hosted HTTP transport is included here.