SkillFoxx

Keploy MCP

Keploy MCP generates API tests from an OpenAPI spec, runs them and lists uncovered endpoints.

  • 19,714 ★ on the repository
  • Last commit
MCP serverOfficial

Extended access

Why this level

  • The Keploy key gives access to the account applications and test suites.
  • Tests run against the given API and send real requests to the staging server.
All reasons and checks

keploy/keployAuthor:keploy

Install

Manual install

Text for your agent

Add the remote MCP `https://api.keploy.io/client/v1/mcp` with the `Authorization: Bearer <key>` header, taking the key from an environment variable. Start with `get_auth_status`, then `generate_and_wait` and `run_and_report`.

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • The Keploy key gives access to the account applications and test suites.
  • Tests run against the given API and send real requests to the staging server.

This is third-party code. Review the repository files before installing.

What it does

The MCP endpoint is built into the Keploy API server at `https://api.keploy.io/client/v1/mcp`. It opens the automated test generation platform to an agent: instead of parsing command-line output, the agent calls typed tools and gets JSON answers. Tools include creating an application, generating test suites from an OpenAPI spec, running suites against a target API, listing suites, recorded real traffic with requests, responses and dependency mocks, and the automatically built schema. The composite calls `generate_and_wait`, `run_and_report` and `get_coverage_gaps` generate tests, run them and return uncovered endpoints with priorities. By default the server runs in tool-search mode: the list shows a few meta-tools and the rest are found through `search_tools`.

Who it is for. Teams with an OpenAPI spec that want tests and coverage reports straight from Claude Code, Cursor or GitHub Copilot.

Good fit when

  • The API has an OpenAPI spec and a test suite is generated from it without manual markup.
  • After a run the agent takes the uncovered endpoints and writes tests aimed exactly at them.
  • Suites run against a staging server and the failed checks are analysed.
  • Recorded real traffic is read together with dependency mocks to understand service behavior.

Not a fit when

  • There is no Keploy account and setting up a platform for a few tests is not worth it.
  • The API key cannot live in the client config: it is sent in the `Authorization` header.
  • You only want code unit tests: this MCP is about API testing.

Example request

Generate tests from my OpenAPI spec, run them against the staging server and show the uncovered endpoints.

MCP

Transport
http
Authentication
API key

Environment variables

Authorization required, secret
Keploy API key as `Bearer kep_...`, issued on the settings page

Limitations

Agent test generation belongs to the Enterprise plan in the documentation, so a Keploy account and an API key from the settings page are required. The key is shown once and sent as `Authorization: Bearer kep_...`, and it must not be stored in a repository. Running tests needs Keploy installed and a live API under test. In tool-search mode the list is short at first: schemas come from `get_tool_schema` and tools are found with `search_tools`. For a self-hosted platform, replace the address with your own API server.

How to disable. Remove the `keploy` entry from the client settings and revoke the API key at `app.keploy.io/settings/api-keys`. Created applications and test suites stay in the Keploy account.

FAQ

Is the server paid?

Agent test generation sits in the Enterprise plan, see the Keploy site for terms. Sign-in uses an API key.

How do I connect Claude Code?

Add a `keploy` server of type `http` with the address `https://api.keploy.io/client/v1/mcp` and the header `Authorization: Bearer kep_<key>` to `~/.claude.json` or the project settings.

Where do I get the API key?

At `app.keploy.io/settings/api-keys`. The key is shown once; copy it together with the word `Bearer`.

Why is the tool list so short?

By default the server runs in search mode: `get_auth_status`, `get_setup_instructions`, `search_tools`, `get_tool_schema` and `invoke_tool` are visible. Other tools are called by name.

Is there a path without MCP?

Yes. The documentation describes a workflow through the `keploy` command line with structured JSON output.

Discussion

No questions or reviews yet.

MCP servers

MCP server: an agent sends GET, POST, PUT, DELETE and PATCH to your REST API and gets back status, headers and body

MCP serverExtended access104

sunpeak tests any MCP server in local replicas of the ChatGPT and Claude hosts and builds MCP Apps with React

CLIExtended access219
Official

The official Testkube MCP server: an agent runs test workflows in Kubernetes and reads run logs, artifacts and metrics.

MCP serverBroad access1.7K
Official

Official Qase MCP server with 42 tools: cases, runs, defects, CI reports and search in the QQL query language

MCP serverBroad access37

MCP Interviewer checks an MCP server before an agent connects: provider limits, tool tests and a Markdown report.

CLIExtended access158

README in short

The README covers the platform and traffic recording and does not mention MCP. Connection and tools are in the agent test generation documentation.

README badge

Report a problem