Hoppy
A command-line tool released by Rebellion Labs that automatically converts existing OpenAPI contracts into MCP servers. It supports dry-run previews of tool definitions and a headless mode for integration with existing backends.
Timeline (last 90 days)
Hoppy is a command-line tool that can directly convert an existing OpenAPI contract into an MCP server without writing server code.
In Hoppy, after running the serve command with the dry-run flag, operations in the OpenAPI contract automatically become MCP tools, read-only operations are marked as read-only, and cancellation operations are marked as destructive.
Hoppy's dry-run mechanism allows developers to see the complete tool definitions that the AI client will receive before formal startup, including input schemas and safety hints, and can estimate context overhead (for example, about 250 tokens).
Hoppy's Headless mode places an MCP endpoint in front of an existing backend, leaving business logic in place, forwarding tool calls to existing APIs, and keeping authentication, validation, and monitoring in effect.
Hoppy's Capability Graph provides business context to the model through contract-level extensions (such as capability, intent, effects, planning facts, client facts, and outcome) without changing the API itself.
In Hoppy's demonstration, for a simulated wholesale backend containing 23 services and 276 operations, if all were exposed to the model, the user would consume about 16,000 tokens before even speaking.
Hoppy's planner never calls APIs directly; it only decides what is safe as the next step, and when necessary facts are missing it declares that clarification is needed rather than fabricating a path.
Hoppy's auditing mechanism verifies the completeness of the vocabulary before the model touches it, and in CI, gaps block the build like a failing test.
In Hoppy, after the same serve command recognizes the Arazo document, it exposes each workflow as an MCP tool; the AI only needs to call fulfillOrder with the order ID, and Hoppy executes the three-step operation against three independent services in sequence and returns the logistics information.
When using Hoppy, users can let the model ask questions before acting by adding a small number of extensions (such as capability, intent, effects, etc.) to the API, thereby obtaining more accurate tool calls.
All Facts (10)
Hoppy is a command-line tool that can directly convert an existing OpenAPI contract into an MCP server without writing server code.
50%UnverifiedIn Hoppy, after running the serve command with the dry-run flag, operations in the OpenAPI contract automatically become MCP tools, read-only operations are marked as read-only, and cancellation operations are marked as destructive.
50%UnverifiedHoppy's dry-run mechanism allows developers to see the complete tool definitions that the AI client will receive before formal startup, including input schemas and safety hints, and can estimate context overhead (for example, about 250 tokens).
50%UnverifiedHoppy's Headless mode places an MCP endpoint in front of an existing backend, leaving business logic in place, forwarding tool calls to existing APIs, and keeping authentication, validation, and monitoring in effect.
50%UnverifiedHoppy's Capability Graph provides business context to the model through contract-level extensions (such as capability, intent, effects, planning facts, client facts, and outcome) without changing the API itself.
50%UnverifiedIn Hoppy's demonstration, for a simulated wholesale backend containing 23 services and 276 operations, if all were exposed to the model, the user would consume about 16,000 tokens before even speaking.
50%UnverifiedHoppy's planner never calls APIs directly; it only decides what is safe as the next step, and when necessary facts are missing it declares that clarification is needed rather than fabricating a path.
50%UnverifiedHoppy's auditing mechanism verifies the completeness of the vocabulary before the model touches it, and in CI, gaps block the build like a failing test.
50%UnverifiedIn Hoppy, after the same serve command recognizes the Arazo document, it exposes each workflow as an MCP tool; the AI only needs to call fulfillOrder with the order ID, and Hoppy executes the three-step operation against three independent services in sequence and returns the logistics information.
50%UnverifiedWhen using Hoppy, users can let the model ask questions before acting by adding a small number of extensions (such as capability, intent, effects, etc.) to the API, thereby obtaining more accurate tool calls.
50%