Steadybit MCP Server

Steadybit MCP Server

The Steadybit MCP server bridges AI assistants such as Claude with the Steadybit resilience testing platform. It exposes a rich set of tools that let language models query, manage, and trigger chaos experiments directly from conversational interfaces. By turning Steadybit’s REST API into a first‑class AI capability, developers can embed automated resilience testing workflows into chatbots, documentation assistants, or continuous‑integration pipelines without writing any custom code.

This server solves the problem of integrating complex resilience operations into natural‑language workflows. Traditional chaos engineering requires navigating a web UI, issuing CLI commands, or scripting API calls—tasks that are cumbersome for non‑technical stakeholders. With the Steadybit MCP, a user can simply ask an AI assistant to “list all experiment designs for the ops team” or “run a template named latency spike in production.” The assistant translates the request into the appropriate tool call, receives structured data back from Steadybit, and presents it in an easy‑to‑read format. This lowers the barrier to experimentation, accelerates incident response drills, and promotes a culture of continuous resilience testing.

Key capabilities are grouped into intuitive tools:

• Experiment Discovery – , and allow quick retrieval of design metadata, enabling assistants to surface available experiments or templates.
• Execution Management – , , and the optional tool let users inspect run histories, filter by state or environment, and launch new experiments on demand.
• Contextual Metadata – Tools such as , , and provide contextual lists that aid in constructing valid requests or troubleshooting failures.
• Scheduling Insight – reveals recurring experiments, helping teams keep their resilience calendar up to date.

Real‑world scenarios include: a DevOps engineer using an AI chat to pull the latest experiment run status before a deployment; a security analyst asking the assistant to trigger a network latency template in a staging environment; or a product manager reviewing all experiments scheduled for the next sprint through natural language queries. In each case, the AI assistant acts as a conversational front‑end that handles authentication, input validation, and error handling automatically.

Integration into AI workflows is seamless. The MCP server follows the Model Context Protocol, so any Claude‑compatible client can register it as a tool set. The server authenticates via an API token, supports environment‑specific filtering, and even allows dynamic capability toggling through environment variables. This design makes it straightforward to enable or disable powerful actions—such as creating experiments from templates—according to organizational policy.

Unique advantages of the Steadybit MCP include its fine‑grained filtering options (e.g., state, environment, date ranges), the ability to create experiments from templates with placeholder substitution, and the optional exposure of scheduled runs. These features give developers precise control while keeping interactions human‑friendly, thereby enabling rapid experimentation, continuous learning, and proactive resilience engineering—all driven by conversational AI.