# TestKase MCP Server: The First AI-Native Test Management Platform What if you could manage your entire test suite by talking to an AI? Not generating test ideas in ChatGPT and copy-pasting them into a spreadsheet. Not asking an AI to write Selenium scripts. Actually managing your test cases, test cycles, execution results, and quality reports through natural conversation — with every action happening in your real test management system, in real time. This is not a hypothetical. This is what TestKase does today. TestKase is the first and only test management platform to ship a Model Context Protocol (MCP) server — the open standard that lets AI agents like Claude, Cursor, GitHub Copilot, and Claude Code talk directly to your test management tools. No browser tabs. No manual data entry. No copy-paste workflows. Just tell the AI what you need, and it does it. TestKase is the only test management tool in the market with MCP server support, a built-in AI agent, and 40+ AI-powered reports. No other TMT — not TestRail, not Zephyr, not Qase — offers agentic test management capabilities. ## What Is MCP and Why Should QA Teams Care? The Model Context Protocol (MCP) is an open standard created by Anthropic that defines how AI agents communicate with external tools and services. If you've used ChatGPT plugins or function calling, MCP is the evolution of that idea — but open, standardized, and supported across multiple AI platforms. Here's the simplest analogy: **MCP is like USB for AI.** Before USB, every device needed its own proprietary connector. USB created one standard that works everywhere. MCP does the same for AI agents — one protocol that lets any AI client talk to any tool. ### Why this matters for QA AI agents are getting smarter. They can reason about requirements, write test cases, analyze failure patterns, and make prioritization decisions. But without a way to connect to your actual test management system, they're limited to generating text that a human has to manually transfer. MCP eliminates the human-in-the-middle. When an AI agent has MCP access to your test management platform, it can: - Create test cases directly in your system (not in a chat window) - Link test cases to test cycles and assign them to team members - Execute tests and record pass/fail/blocked results - Pull real-time reports and analyze quality trends - Make decisions based on actual data, not hallucinated metrics The catch? Your test management tool needs to support MCP. And right now, only one does. ## What TestKase's MCP Server Can Do TestKase's MCP server exposes **11 tools** that cover the entire test management lifecycle. Each tool supports multiple actions, giving AI agents fine-grained control over your test suite. ### Test Case Management in Detail The `manage_testcase` tool alone replaces hours of manual work. When you tell an AI agent to "create a login test case with high priority and three test steps," here's what happens behind the scenes: 1. The AI calls `get_project_structure` to discover allowed values for priority, status, and available folders 2. It calls `manage_testcase` with action `create`, passing the title, priority, summary, and test steps 3. TestKase creates the test case, adds each test step, and returns the ID and URL 4. The AI confirms: "Created TEST-142: Login with valid credentials (High priority, 3 test steps). View it here: [link]" For bulk operations, the `create_bulk` action accepts an array of test cases — create 10, 20, or 50 test cases in a single call: ``` "Create 10 test cases for the checkout flow covering: happy path, empty cart, expired coupon, international shipping, payment failure, session timeout, back button, guest checkout, quantity limits, and inventory out-of-stock" ``` The AI generates all 10 with appropriate titles, priorities, preconditions, and test steps — then creates them in your project in one batch. ### Test Cycle Management: 8 Actions Test cycles are where the rubber meets the road. The `manage_test_cycle` tool supports eight distinct actions: 1. **create** — Create a new cycle with title, summary, status, date range, and folder 2. **update** — Modify any cycle field 3. **delete** — Remove one or more cycles 4. **get_details** — Get complete cycle information with execution summary 5. **get_testcases** — List all test cases in the cycle with search and pagination 6. **link_testcases** — Add existing test cases to the cycle 7. **unlink_testcases** — Remove test cases from the cycle 8. **assign_testcases** — Assign linked test cases to a specific team member A single conversation can chain these actions: ``` "Create a Sprint 14 regression cycle, link all critical and high priority test cases, and assign the authentication ones to Sarah" ``` The AI makes 3-4 tool calls in sequence: create the cycle, search for critical/high test cases, link them, search for auth-related ones, and assign them to Sarah. ### Test Execution The `execute_tests` tool records real test execution results — individually or in bulk: ```json { "testcase_id": "TEST-142", "execution_status": "pass", "actual_result": "Login successful. Dashboard loaded in 1.2s.", "environment": "staging" } ``` Status options are `pass`, `fail`, `blocked`, and `not_executed`. Bulk execution lets you record results for an entire cycle in one call — useful when importing results from automation frameworks. ## 40+ Reports via Natural Language This is where TestKase's MCP server truly differentiates. Instead of navigating dashboard menus and configuring filters, just ask: > "Show me the execution summary for the Sprint 14 cycle" > "What's our requirement coverage for the payment module?" > "Which testers have the highest workload this sprint?" > "Are we ready to release?" The `get_report` tool supports **40+ report types** across seven categories: ### AI Insights: Reports That Think The AI insight reports go beyond raw data. They use machine learning to analyze patterns in your test history: - **Predictive failure** — Which test cases are most likely to fail in the next cycle, based on historical patterns - **Smart prioritization** — AI-ranked test execution order that maximizes defect discovery per hour - **Flaky test detection** — Tests that pass and fail inconsistently, eroding confidence in your suite - **Suite optimization** — Identifies redundant test cases, coverage gaps, and opportunities to reduce suite size without losing coverage - **Release readiness** — A composite score that weighs pass rates, open defects, coverage gaps, and risk factors to answer "are we ready to ship?" These aren't generic dashboards. They're actionable recommendations generated from your actual test data. ## Real-World Workflows ### Workflow 1: Sprint Planning QA Setup **The manual way:** Open TestKase, create a test cycle, navigate to test cases, filter by module, select relevant cases, link them, assign testers, update the cycle status. 15-20 minutes of clicking. **The AI way:** ``` "Create a 'Sprint 14 Regression' cycle for PRJ-1001. Link all test cases from the Authentication and Payment folders. Assign auth tests to Sarah and payment tests to Daniel. Set the cycle dates from March 25 to March 28." ``` Done in one conversation. Four tool calls. Under 30 seconds. ### Workflow 2: Test Case Creation from Requirements **The manual way:** Read the requirements document, open the test case creation form, type each test case with steps, set priorities, assign labels, choose a folder. Repeat for every requirement. **The AI way:** ``` "Read these requirements and create test cases: 1. Users must be able to reset their password via email 2. Reset links expire after 24 hours 3. Links can only be used once 4. Rate limit: max 3 reset requests per hour Create them in the Authentication folder with High priority and include test steps for happy path and edge cases." ``` The AI generates 8-12 test cases covering happy paths, edge cases, boundary conditions, and security scenarios — each with detailed test steps, preconditions, and expected results. ### Workflow 3: Release Readiness Check **The manual way:** Open the dashboard, navigate to reports, check execution summary, check coverage, check open defects, check trends, synthesize the information, write a summary for stakeholders. **The AI way:** ``` "What's our release readiness for PRJ-1001? Check the Sprint 14 cycle results, requirement coverage, and any open critical defects." ``` The AI pulls three reports (release readiness, execution summary, defect hotspots), analyzes the data, and responds with: > "Sprint 14 cycle shows 94% pass rate with 2 failed tests in the Payment module. Requirement coverage is 87% — the Refund module has 3 uncovered requirements. No critical defects are open. One high-priority defect (DEF-234) is in progress. Release readiness score: 78/100. Recommendation: address the 2 payment failures and the Refund coverage gap before release." ### Workflow 4: End-of-Day Status Update ``` "Give me a summary of today's test execution across all active cycles. Include pass rates, new failures, and who executed the most tests." ``` The AI pulls execution-by-cycle, execution-by-tester, and trend reports, then generates a concise status update you can paste into Slack. ## Works With Your Favorite AI Tools TestKase's MCP server uses the standard stdio transport, making it compatible with any MCP client. Here's how to set it up: ### Claude Desktop Add to your `claude_desktop_config.json`: ```json { "mcpServers": { "testkase": { "command": "npx", "args": ["-y", "testkase-mcp-server@latest"], "env": { "TESTKASE_PAT_TOKEN": "xyz_your_token_here" } } } } ``` ### Cursor IDE Add to `.cursor/mcp.json` in your project root: ```json { "mcpServers": { "testkase": { "command": "npx", "args": ["-y", "testkase-mcp-server@latest"], "env": { "TESTKASE_PAT_TOKEN": "xyz_your_token_here" } } } } ``` ### GitHub Copilot Add to VS Code settings (`.vscode/mcp.json`): ```json { "servers": { "testkase": { "command": "npx", "args": ["-y", "testkase-mcp-server@latest"], "env": { "TESTKASE_PAT_TOKEN": "xyz_your_token_here" } } } } ``` ### Claude Code CLI ```bash claude mcp add testkase -- npx -y testkase-mcp-server@latest ``` That's it. One line. Your Claude Code session now has full access to your test management system. From zero to connected in under 2 minutes. Generate a PAT from your TestKase dashboard (My Profile → API Keys), add the config, and start talking to your test suite. ## Built-in AI Agent: Zero-Setup Alternative Not everyone uses Claude Desktop or Cursor. For teams that want AI-powered test management without any local setup, TestKase includes a **built-in AI agent** directly in the dashboard. Press **Ctrl+K** (or click the AI icon) and a conversational sidebar opens. It has the same 11 tools as the MCP server — the same capabilities, the same natural language interface — but runs entirely in the browser. ### Key features of the built-in agent: - **Same 11 tools** as the MCP server — no capability gap - **Real-time tool tracking** — see which tools the AI is calling, their status (running/completed/failed), and results - **Streaming responses** — watch the AI think and respond in real time - **Conversation history** — pick up where you left off (persisted for 24 hours) - **Multi-LLM support** — powered by Claude, OpenAI, or Gemini depending on configuration - **No setup required** — works instantly for any TestKase user The built-in agent is ideal for: - Teams that want AI test management without configuring local tools - Quick questions ("How many test cases do we have in the Payment folder?") - Non-technical team members who want to pull reports or check status - Demo and evaluation — try AI-powered test management before committing to a workflow ## Why No Other Test Management Tool Has This Let's be direct about the competitive landscape: This is not a marginal advantage. It's a category difference. Other TMTs are form-based CRUD applications. TestKase is an AI-native platform that happens to also have a UI. ### Why the gap exists Building an MCP server requires: 1. **A well-designed API** — Every operation must be expressible as a structured tool call 2. **Smart field coercion** — AI agents send values like "high" or "High" or "HIGH" — the server must normalize them 3. **Error recovery** — When a tool call fails, the server returns actionable hints (not generic 500 errors) 4. **Context-aware design** — The tool schema must guide the AI toward valid workflows (discover field values before creating) 5. **Security** — PAT-based auth with the same permission model as the web app This is months of engineering work on top of an already-complete platform. Legacy TMTs would need to rebuild their entire API layer. TestKase was designed with AI integration as a first-class concern from day one. ## Security and Trust AI agents operating on your test data is a legitimate security concern. Here's how TestKase addresses it: - **Personal Access Tokens (PAT)** — No passwords shared with AI tools. Tokens can be revoked instantly from the dashboard. - **Same permissions model** — The MCP server operates with your user's exact permissions. If you can't delete a project in the UI, you can't delete it via MCP. - **HTTPS only** — All API communication is encrypted in transit. The MCP server connects to `https://api.testkase.com`. - **Stateless tools** — No test data is cached or stored by the AI agent or MCP server. Each tool call is an independent, authenticated API request. - **Audit trail** — Every action taken via MCP is logged the same way as UI actions. You can see who (which PAT) created, modified, or deleted any resource. - **Rate limiting** — API rate limits apply equally to MCP calls, preventing runaway automation. Create a dedicated PAT for each AI tool you connect. This way, you can see exactly which tool performed which actions in the audit log, and revoke access per-tool if needed. ## The Future of QA Is Agentic The role of the QA engineer is changing. Not disappearing — evolving. Here's the shift: **Before AI agents:** - QA engineers spend 30-40% of their time on manual data entry — creating test cases, updating statuses, writing reports - Context switching between tools (Jira, test management, CI/CD, Slack) fragments focus - Reports are generated weekly because pulling them is tedious - Scaling QA means hiring more people **With AI agents + TestKase MCP:** - Test case creation is conversational — describe what you want, AI creates it with steps and metadata - Multi-step workflows (create cycle → link cases → assign → set dates) happen in one conversation - Reports are pulled on demand — "What's our coverage?" is a 5-second question, not a 15-minute dashboard dive - QA engineers focus on strategy, exploratory testing, and edge cases — the work that actually requires human judgment The QA engineer of 2026 is a **test architect** — someone who designs testing strategy, directs AI agents, reviews AI-generated test cases, and focuses on the creative, high-value work that machines can't do. TestKase is the platform built for that future. ## Getting Started 1. **Sign up** at [testkase.com](https://www.testkase.com/signup) — free for up to 3 users 2. **Generate a PAT** from My Profile → API Keys 3. **Connect your AI tool** using the config snippets above 4. **Start with a simple command:** "List my projects" or "Show me the project structure for PRJ-1001" 5. **Build from there:** Create test cases, set up cycles, execute tests, pull reports — all through conversation The entire setup takes under 2 minutes. No credit card required. No sales call needed. Try TestKase Free — AI-Native Test Management → If your test management tool can't talk to AI agents, it's already behind. The question isn't whether agentic testing will become the standard — it's whether you'll adopt it now or spend the next year doing manually what AI could do in seconds. TestKase is ready. Is your test management tool?

Blog

TestKase MCP Server: The First AI-Native Test Management Platform