目录 / QualityMax MCP
MCP
鉴权未知
未评级
已上架
QualityMax MCP
AI-native test automation for web apps. Create and manage projects, test cases, automation scripts, test runs, and QualityMax workflows through secure OAuth.
该来源不提供完整文件导出（国内平台多为平台内托管），仅存元数据与原链
接入信息
- 传输形态
- http
- 鉴权方式
- 鉴权未知
- 端点
https://qualitymax-mcp--qualitymax.run.tools
鉴权方式未标注，请核对官方文档后再接入——不要直接使用以下片段
{
"mcpServers": {
"QualityMax MCP": {
"url": "https://qualitymax-mcp--qualitymax.run.tools"
}
}
}
能力清单
| 工具 | 说明 |
|---|---|
| import_repository | Import and analyze a GitHub repository for testing. Creates a default project if none specified. |
| generate_tests | Generate Playwright tests for a specific feature or component |
| run_tests | Execute Playwright tests or verified generated Appium mobile flows and return a compact per-script summary (status, duration, failure reason, next action). Managed mobile runs are asynchronous; poll get_execution using the returned execution_id. Full stdout/stderr, traces, and artifact paths are opt-in — request them only when the compact failure is not enough to act on. |
| apply_patch_and_run | Patch an automation script and immediately re-run it, returning one compact result. Use this instead of update_script followed by run_tests: it collapses the fix-and-verify loop into a single round trip. |
| analyze_coverage | Analyze test coverage and identify gaps in the repository |
| generate_area_plan | Run deep gap analysis on a repository to discover and map testing areas. |
| generate_tests_for_areas | Generate test cases for specific discovered test areas. |
| get_discovery_graph | Return the repository's evidence-grounded discovery graph: a lineage of repository → evidence → test areas → test cases → automation scripts, with grounding status and confidence per node. Use to audit test coverage lineage. |
| ai_review | Perform AI-powered code review and suggest improvements |
| get_review_preferences | Get the user's AI code-review preferences (which categories to check, which to skip). Returns effective preferences: per-repo overrides merged over global defaults. Call without repository_id for global defaults. |
| set_review_preferences | Set or update AI code-review preferences. Controls which categories the reviewer checks (security, performance, test_coverage, type_safety, accessibility, style, secrets_scanning, ai_safety_for_agents) and an optional custom_focus string. Use scope='global' for defaults that apply everywhere, or scope='repo' with a repository_id for per-repo overrides. |
| create_pr | Create a pull request with generated tests or fixes |
| whoami | Return identity of the user owning the current API token: user_id, email, username, full_name, role, tier, and a count of accessible projects. Use this first when debugging 'Access denied' errors to confirm which account the token belongs to. |
| list_projects | List projects for the current user, including each project's vanity URL slug and project type (web or native_mobile). ALWAYS call this before creating a new project to check if a similar project already exists. |
| create_project | Create a new project (optionally with main_url). Will check for similar existing projects and ask for confirmation if found. |
| update_project | Update an existing project (e.g., set main_url for AI crawl) |
| get_project_memory | Return compact durable memory from prior MCP agent negotiations, project learnings, and verified test runs. Call this before generating or repairing tests for an existing project; then call get_script only for the most relevant verified script to keep context and cost low. |
| get_repository | Get repository details and stored analysis |
| refresh_repository | Refresh an existing repository analysis at its latest commit. Private repositories reuse the stored GitHub App connection. |
| list_repositories | List repositories for a project |
| repo_quality | Get the imported repository's stored quality signal without starting a new scan. |
| repo_coverage | Get the imported repository's stored coverage signal without starting a new scan. |
| list_test_cases | List test cases for a project |
| create_test_case | Create a test case in a project. If no steps are provided, detailed steps are auto-generated from the title/description (disable with auto_generate_steps=false). |
| import_test_cases_from_document | Import test cases from document text content. The AI will analyze the text and extract test cases automatically. |
| import_from_testrail | Import test cases from TestRail into a project (requires saved TestRail credentials) |
| start_ai_crawl | Start an AI crawl job for a project (requires project.main_url unless url is provided) |
| start_ai_crawl_from_test_case | Start an AI crawl job using an existing test case as instructions. IMPORTANT: This tool will first generate a virtual test plan that MUST be approved by the user before proceeding. You must present the plan to the user, wait for their approval, then call this tool again with the approved_virtual_test_plan parameter to proceed with the crawl. |
| check_ai_crawl_status | Check the status of an AI crawl job. Use this to see if a crawl has completed and inform the user when tests are ready. |
| start_mobile_audit | Start a detailed Mobile Readiness Audit for a project. It captures responsive evidence at 1440px, 768px, and 375px, checks console/link/accessibility/layout risks, and prepares a shareable report. Uses the project main_url unless url is provided. |
| get_mobile_audit_status | Get status, score, and report path for a Mobile Readiness Audit. |
| list_mobile_audits | List Mobile Readiness Audits for a project, newest first. |
| share_mobile_audit | Create or return the public share link for a completed Mobile Readiness Audit. |
| parse_mobile_audit_report | Convert extracted text from a QualityMax-style mobile audit PDF or HTML report into structured findings: ID, severity, surface, breakpoints, root cause, and fix prompt. Use the result to create or link follow-up work; this tool never changes project data. |
| get_mobile_audit_evidence | Return the screenshot URLs and bounded scanner evidence for a completed Mobile Readiness Audit. Use this before Claude writes a qualitative report. |
| render_mobile_audit_report | Store a structured BYO-Claude mobile-audit report and render a shareable HTML report plus a downloadable PDF. This does not call a QualityMax model. |
| import_document_and_crawl | Complete workflow: Import document/text, create test cases, generate virtual test plan for approval, then initiate AI crawl. IMPORTANT: This tool will first generate a virtual test plan that MUST be approved by the user before proceeding. You must present the plan to the user, wait for their approval, then call this tool again with the approved_virtual_test_plan parameter to proceed with the crawl. |
| show_available_commands | List all available commands and tools that the assistant can use. Use this when the user asks 'what can you do?' or 'list your capabilities'. |
| get_project_summary | Get a summary of a project including counts of test cases, scripts, and executions. Use this when the user asks 'what's in my project?' or 'show me project stats'. |
| get_test_case | Get detailed information about a specific test case including its description, steps, and related automation scripts. Pass `if_version_matches` to skip the body when nothing changed, and `fields` to fetch only what you need. |
| get_script | Get an automation script's details including the generated Playwright/Cypress test code. Pass `if_version_matches` with the version from a previous read to get a cheap `not_modified` answer, and `fields` to fetch only what you need. |
| list_scripts | List automation scripts for a project or test case. Returns script names and IDs (use get_script to view the code). |
| get_execution | Get a test execution's status, timing, and outcome. Log entries and artifact paths are counted by default and returned only when explicitly requested. |
| search_test_cases | Search for test cases by title, description, or tags. Better than listing all test cases when looking for something specific. |
| query_data | Query data from the system using flexible queries. Use this for exploratory queries or when you need to list data with filters. For specific single-item lookups, prefer specialized tools like get_test_case, get_execution, etc. |
| parse_qtml | Parse a QTML (Quality Testing Meta Language) source string and return its structured representation as JSON. Use this to inspect the contents of a QTML document. |
| validate_qtml | Validate a QTML source string. Returns whether it is valid and any errors found. |
| export_project_as_qtml | Export a QualityMax project and its test cases as a QTML document string. |
| import_qtml | Import a QTML source string into QualityMax, creating a project and/or test cases from the QTML contract and intents. |
| convert_qtml_to_playwright | Import a QTML source string and then trigger AI Playwright code generation for each imported test case. |
| test_deployed_environment | Test a deployed environment end-to-end. Discovers the UI, generates a Playwright test, and executes it. The run is ephemeral unless project_id is provided, in which case a linked test case, script, and result are saved. Returns pass/fail with test code and screenshots. |
| get_ai_crawl_results | Get the results of a completed AI crawl job including generated test scripts, test cases, and flow memory. |
| generate_gap_tests | Generate test cases to fill coverage gaps found by analyze_coverage. Analyzes the repository's deep-scan results and generates E2E, smoke, integration, and security test cases. |
| generate_code_for_test_case | Generate automation test code for a specific test case using AI. Supports Playwright, pytest, Cypress, Selenium, Vitest, Jest, Kotlin, Scala, Go (`go test`), and Rust (`cargo test`). When `framework` is omitted, the framework is auto-resolved from project settings, linked-repository language detection, and the test category. |
| check_generation_status | Check a background code-generation job started by generate_code_for_test_case (personal Codex / Claude Code seat routes return a job_id instead of code, because a seat run takes minutes). Poll about every 30s. A seat runs one job at a time, so wait for a terminal status ('completed', 'failed', or 'timed_out') before generating the next test case rather than starting several in parallel. |
| update_test_case | Update an existing test case's fields (title, description, steps, expected result, category, priority). |
| delete_test_case | Delete a test case. Requires ownership of the parent project. |
| delete_project | Delete a project. Requires ownership. This will also delete all associated test cases, scripts, and executions. |
| update_script | Update an automation script's name or code. Prefer `patch` over `code`: it sends only the lines you are changing instead of re-uploading the whole script, which is the single largest token saving available on this server. |
| update_mobile_flow | Update a managed mobile script from its canonical QTML mobile flow. Validates the flow and platform, rejects stale revisions, increments the revision, and deterministically recompiles the executable script. Use this instead of update_script for managed mobile flows. |
| delete_script | Delete an automation script. Requires ownership of the parent project. |
| list_ai_crawl_jobs | List AI crawl jobs for a project, ordered by most recent first. |
| enhance_test_case | Enhance a test case with AI — adds more detailed steps, verification points, and improves the description. |
| trigger_framework_run | Trigger a full framework test execution for a project. The framework must already exist (tests added via accept_pending_tests or add_to_framework). Supports every registered framework — playwright, pytest, cypress, rust_cargo (cargo test), go_test (go test -json), k6, etc. The runner is picked from the project's framework type; native (Rust/Go) runs execute via cargo/go and return normalized console_logs just like browser runs. |
| add_script_to_framework | Add an existing automation script to the project's test framework. Fetches the script by ID, ensures the framework exists (creating it if needed), and writes the test file into the framework directory. |
| export_framework | Export a project's test framework as a zip archive. Since MCP cannot return binary data, the zip is saved to a temporary file and the file path and size are returned. |
| get_install_command | Get the install command for the QualityMax local agent. Returns a pip install command with the user's registration secret. |
| run_native_test | Run a Rust (`cargo test`) or Go (`go test -json ./...`) automation script and return a normalized result with console_logs, passed/failed/total, stdout, stderr, and a duration. Use this when you have a compiled-language script (framework: rust_cargo or go_test) and want the native runner path. For Playwright / pytest scripts, use run_tests instead. |
| setup_cicd | Create a Pull Request on the linked GitHub repository that adds a GitHub Actions workflow file running the project's test suite. Auto-detects the framework from the repo's analyzed languages (playwright / pytest / go / rust) when omitted, and for Rust auto-detects apt packages from Cargo.lock (glib-sys→libglib2.0-dev, openssl-sys→libssl-dev, libxdo-sys→libxdo-dev, etc.) so the generated workflow builds on a bare ubuntu-latest runner. Requires the GitHub App to be installed on the target repo. Returns the PR URL, PR number, workflow file path, detected framework, and the apt packages that were injected. |
| accept_pending_tests | Accept pending test proposals for a project, writing them into the framework. Can accept a single test or all proposed tests. |
| run_agent_eval | Run a conversation evaluation against an AI agent endpoint. Sends test case prompts to the agent, judges responses on accuracy/helpfulness/safety/relevance/conciseness. Returns per-test scores and overall pass rate. |
| run_adversarial_eval | Run adversarial safety testing against an AI agent. Sends 39 attack prompts (prompt injection, jailbreak, data extraction, PII leakage, bias probing, off-topic) and reports which attacks succeeded vs were blocked. |
| get_agent_eval_results | Get detailed results for an agent evaluation run, including per-test scores, judge reasoning, and conversation transcripts. |
| list_adversarial_prompts | Browse the adversarial prompt library. 39 curated attack prompts across 6 categories for testing AI agent safety. |
| list_end_user_personas | List every built-in Agentic Eyes end-user persona plus the authenticated user's saved custom personas. Each result includes a runnable_id that can be passed directly as the persona argument to run_persona_review or run_persona_consensus. |
| run_persona_review | Browse a live website as a customizable AI end-user persona (e.g. Sally, the Dreaming Planner, or the Worst-Customer-Ever) and return a plain-language 'message from user' UX report: what they tried, where they got stuck, what confused them, the questions they had, whether they'd convert, and the top fixes. With drive=true the persona actually clicks through a multi-step flow (not just the landing page) and reports the journey it took. Part of Agentic Eyes. |
| export_agentic_eyes_report | Generate a standalone HTML or PDF report from a saved, project-attributed Agentic Eyes persona review. HTML is returned as text; PDF is returned as base64 so MCP clients can save it without binary corruption. |
| run_persona_consensus | Run the SAME persona review across several different AI models and rank the findings by how many models independently reported them. Findings that agree across models are high-confidence ('4/4 models flagged no pricing'); findings only one model saw are exploratory and listed in a divergence section. Use this when a finding is going in front of a customer or driving a fix, and a single model's word isn't enough. Costs roughly N times a single review. Part of Agentic Eyes. |
| k6_create_script | Create a k6 performance test script. Supports load, stress, spike, soak, smoke, breakpoint, API, security, and browser test types. |
| k6_list_scripts | List k6 performance test scripts for a project |
| k6_get_script | Get a k6 script by ID including its code |
| k6_update_script | Update a k6 script's code, name, or configuration |
| k6_run_test | Execute a k6 performance test. Returns an execution ID for polling status. Supports overriding VUs, duration, and thresholds. |
| k6_check_status | Check the status and results of a k6 test execution. Poll this after k6_run_test. |
| k6_generate | Generate a k6 performance test script from a target URL and optional endpoint list. Auto-creates load profiles with thresholds. |
| k6_generate_from_test_case | Generate a k6 performance test from an existing QualityMax test case. Extracts HTTP interactions and builds a load test. |
| jmeter_to_k6 | Convert a JMeter .jmx test plan to k6 JavaScript. Handles thread groups, HTTP samplers, timers, assertions, and extractors. |
| gatling_to_k6 | Convert a Gatling Scala simulation to k6 JavaScript. Handles HTTP builders, injection profiles, checks, and feeders. |
| locust_to_k6 | Convert a Locust Python load test to k6 JavaScript. Handles HttpUser tasks, wait times, on_start/on_stop, and task weights. |
| playwright_to_k6 | Convert Playwright API tests or page navigations to k6 load tests. Extracts HTTP calls and builds k6 equivalents. |
| k6_report | Generate a performance test report. Supports single execution reports, comparison reports, and project-level summaries. |
| max_advice | Proactive advice from Max from QualityMax — the platform's always-on companion. Returns your pending and recent proactive notes (things Max noticed about your projects, test runs, and crawls). Pass refresh=true to have Max look at your current platform state right now and produce a fresh note if there is something genuinely worth saying. Requires Max to be enabled for MCP by an owner and your own stored consent. |
| linear_get_issue | Read one Linear issue by its identifier (e.g. 'ENG-123') or UUID. Returns title, description, state, assignee, priority, labels and URL. Use this whenever the task mentions a Linear issue — do not rely on details pasted into the prompt, which may be stale. Read-only. |
| linear_search_issues | Search the connected Linear workspace. All filters are optional and combine: full-text `query`, `team_key` (e.g. 'ENG'), `assigned_to_me`. Completed and cancelled issues are excluded unless include_completed is true. Read-only. |
| linear_get_comments | Read the comment thread on a Linear issue, oldest first. Use this for the discussion and decisions that are not in the issue description. Read-only. |
| linear_add_comment | Post a comment on a Linear issue as the connected user. Use this to report what a run did — findings, a PR link, why something was not done. WRITE: requires the user to have enabled Linear writes for this run, and only works on issues in the workspace they connected. If it is refused, say so and continue; the refusal cannot be worked around. |
| linear_set_issue_state | Move a Linear issue to a workflow state by name (e.g. 'In Review', 'Done'). Call it WITHOUT `state` first to see the states that issue's team actually allows — state names are per-team and cannot be guessed. WRITE: requires the user to have enabled Linear writes for this run, and only works on issues in the workspace they connected. |
| list_deploy_integrations | List deploy-provider webhooks (Vercel, Netlify, manual) configured for a project. Read-only. The webhook_secret is masked in the response. |
| create_deploy_integration | Create a deploy-provider webhook integration for a project. Auto-generates a webhook URL and secret. Returns the full secret (only shown once), the URL, and provider-specific setup instructions. One integration per provider per project. |
| delete_deploy_integration | Delete a deploy-provider webhook integration. The provider must be told to stop sending events afterwards — this only removes the QualityMax side. |
| test_deploy_integration | Record a test-ping against a deploy integration to mark it active without waiting for a real deploy. Useful to confirm the integration is wired up after pasting the webhook into Vercel/Netlify. |
| update_project_pipeline_gates | Toggle PR-pipeline gate flags on a project. Gates live in project.test_frameworks and control which checks run on each pull request. Only the known gate keys are accepted; all other settings (model config, etc.) are preserved. Pass a gate as true/false. Credit-spending gates (ai_review, sast, bola) default OFF when absent; the rest default ON. |
纠错与举报（发现条目失效、署名有误或涉及侵权？）
提交举报 / 纠错
侵权举报经核验成立后，我们会即时下线该条目并删除已存的内容副本。