← Back to home
SEE IT IN A WORKFLOW
Each familiar chore becomes a contract-driven flow.
Pick a task below. Powerduck runs it from the OpenAPI file on your machine — environments, tests and mocks included, with no account or upload required.
Find your use case
01 Generate docs from the project.
02 Implement from the actual API contract.
03 Run the whole business scenario.
04 Inspect seven protocols in one place.
05 Investigate with real request context.
06 Publish docs and MCP from one API.
01
Generate docs from the project.
Do you still read the source code to find out whether the API docs are right?
WITH POWERDUCK
Scan a supported project from your coding agent through MCP. Generate an API definition and open it as documentation.
WHAT YOU GET
A reviewable API definition + a local docs link.
Ask your agent
Scan this project and give me an API documentation link.
Good to know
Supported frameworks only. Review inferred fields and scan warnings. Local links work on the MCP host while its session is running.
See setup & requirements
↗
02
Implement from the actual API contract.
Have you given AI the spec, only to get different fields and paths back?
WITH POWERDUCK
Connect your coding agent to Powerduck’s MCP. Let it read the actual paths, schemas and auth before implementing, then verify the running API.
WHAT YOU GET
Implementation guided by the contract, checked against real responses.
Ask your agent
Implement this API from its definition, then run contract tests.
Good to know
Checks cover the operations, inputs and assertions you run. They do not prove every behavior of the backend is correct.
See setup & requirements
↗
03
Run the whole business scenario.
Do you manually repeat the same business flow after every API change?
WITH POWERDUCK
Describe a flow such as sign in → create an order → retrieve it. Let AI compose the scenario, pass values between steps and execute the test.
WHAT YOU GET
Step-by-step results and failed assertions, not just a generated test script.
Ask your agent
Test sign in → create an order → retrieve it. Show which step fails.
Good to know
Checks cover the operations, inputs and assertions you run. They do not prove every behavior of the backend is correct.
See setup & requirements
↗
04
Inspect seven protocols in one place.
Do you switch tools just because the next endpoint uses a different protocol?
WITH POWERDUCK
Debug HTTP, GraphQL, WebSocket, SSE, gRPC, MCP and A2A. Inspect responses, streams and tool calls in the same workspace.
WHAT YOU GET
Manual control when you need to inspect exactly what happened.
Good to know
Protocol support differs between desktop and browser. Use the desktop app for native transports such as gRPC.
See setup & requirements
↗
05
Investigate with real request context.
Do you copy requests and responses into chat every time you need AI’s help?
WITH POWERDUCK
Give the assistant the relevant API context. Ask it to send a request, inspect the response and help narrow down the failure.
WHAT YOU GET
A diagnosis grounded in request results, with changes you can review.
Ask your agent
Investigate why this request fails. Check the response before suggesting a fix.
Good to know
AI features require a configured model. Context sent to a remote provider leaves your device; manual editing does not require AI.
See setup & requirements
↗
06
Publish docs and MCP from one API.
Do you maintain a separate hosting setup just to share API docs and MCP?
WITH POWERDUCK
Publish a version of your API to Powerduck Cloud. Give people readable docs and give AI clients an MCP endpoint.
WHAT YOU GET
Hosted documentation and an MCP URL, without running your own hosting stack.
Good to know
Cloud hosting uses your account quota. The uploaded API definition may contain examples or internal details: review it before sharing.
See setup & requirements
↗
Try it on one API first.
Open a specification you already have, or start with the built-in example. Explore the workflow before connecting your own project.