Skip to main content

TestPlanIt MCP Server

The TestPlanIt MCP server lets AI agents — Claude Desktop, Cursor, custom MCP-aware clients — connect to a TestPlanIt instance and query test management data on your behalf. It speaks the Model Context Protocol over stdio JSON-RPC and authenticates via a TestPlanIt API token.

What an agent can ask

  • List, fetch, create, update, and soft-delete test cases (with steps, custom fields, tags, folder breadcrumb, linked issues, and linked automated tests inline)
  • List and create test runs, add cases to existing runs, submit test results, and update run state
  • List sessions, session results, and session findings — the exploratory testing surface — and create or update sessions
  • Create and update milestones, mark them started or complete, and list milestone progress with pooled status rollups inline
  • Resolve issues by external key (Jira / GitHub / Azure DevOps) and walk the issue → linked test cases graph
  • Traverse the bridge between automated tests (repository cases) and manual test cases via repository case links
  • List code repositories configured in a project (with credentials never returned)
  • List folders and tags scoped to a project, with usage counts and tree relationships preserved

See the npm package README for the full tool reference, including request/response schemas for all 42 tools.

Installation

npx @testplanit/mcp-server

There is no daemon to manage and no port to forward. Your MCP-aware client launches the server as a stdio subprocess on demand. The server exits after the client disconnects.

Environment variables

Required

VariableDescription
TESTPLANIT_API_TOKENAPI token from your TestPlanIt profile. Must start with tpi_. Mint one under Profile → API Tokens.
TESTPLANIT_API_URLBase URL of your TestPlanIt instance (e.g. https://yourcompany.testplanit.com).

The server validates TESTPLANIT_API_TOKEN on startup and exits cleanly with a human-readable error if the token is invalid, expired, or unreachable — your agent reports a clean failure rather than hanging.

Next steps

  • Configuration — Claude Desktop / Cursor snippets, token scopes, and the read-only agent token walkthrough
  • Example prompts — agent prompts for issue lookup, run history, and maintenance flows
  • npm package README — full tool catalog with request/response schemas