Skip to main content
Glama
Allure TestOps MCP

Allure TestOps MCP

by iampopovich

Allure TestOps MCP

npm version monthly downloads total downloads package size MIT License CI

Production-ready MCP server for Allure TestOps focused on test cases, launches, test results, and test plans.

Features

  • JWT bearer auth exchange from user API token, with in-memory caching and refresh on expiry.

  • Project-aware tools with optional default project via ALLURE_PROJECT_ID.

  • Project resolution by projectId or projectName for project-scoped operations.

  • Test case custom field support, including lookup and updates.

  • stdio transport for local MCP clients (npx or local build).

Related MCP server: Lucius

Tool Coverage

  • Test cases: list, search, get, create, update, delete, restore, overview, history, scenario, steps (get/update), tags (get/set/bulk-add/bulk-remove), issues (get/set), external links bulk add, custom fields (list/get/set), attachments (list/upload/download/delete)

  • Launches: list, search, get, create, update, delete, close, reopen, statistics, progress, add test cases/plans

  • Test results: list, search, get, create, update, history, assign, resolve, retries, attachments (list/download)

  • Test plans: list, get, create, update, delete, run

  • Analytics: automation chart, group by automation/status, launch duration histogram, mute trend, statistic trend, test case last result, test case success rate

  • Dashboards: list, get, create, update, delete, copy, widget data

  • Defects: list, get, create, update, get test results/cases/launches, get launch defects, find similar failures, link to results (bulk), bulk close/reopen, link/unlink issue, apply matchers

  • Mutes: list, create, delete, get muted test cases, get launch muted results, mute/unmute result, bulk mute/unmute

  • Members: suggest users, list project members, launch member stats, test case members, test result members

  • Shared Steps: list, get, get steps (inline scenario), usage, create, update, archive, unarchive

  • Environments: list env vars, suggest env vars, list schemas, list values, suggest values, get result env vars

User Usage Docs

User-focused runtime and integration docs are in docs/usages/:

Tool Catalog (GitHub Pages)

The auto-generated MCP tool catalog is published via GitHub Pages:

The page content is generated from source tool definitions. When a new tool is added to src/tools/*.ts, the Pages workflow regenerates docs/tools.json and updates the site automatically.

Authentication

This server follows the Allure TestOps API guide:

  1. Use your user-generated API token in ALLURE_TOKEN.

  2. Server exchanges it at /api/uaa/oauth/token.

  3. Received bearer JWT is cached and reused until near expiry.

Reference: https://docs.qameta.io/allure-testops/advanced/api/

Environment Variables

ALLURE_TESTOPS_URL=https://allure-testops.instance.com/
ALLURE_TOKEN=your-api-token
# Optional default project:
# ALLURE_PROJECT_ID=37
  • ALLURE_TESTOPS_URL required

  • ALLURE_TOKEN required

  • ALLURE_PROJECT_ID optional

If ALLURE_PROJECT_ID is not set, tools that require project scope must receive:

  • projectId, or

  • projectName (resolved via /api/project/suggest)

Run Locally

Prerequisites

  • Node.js v18+ and npm

Step-by-step

  1. Clone the repository:

git clone https://github.com/iampopovich/allure-testops-mcp.git
cd allure-testops-mcp
  1. Install dependencies:

npm install
  1. Create your local environment file from the example and fill required values:

cp .env.example .env

Set:

  • ALLURE_TESTOPS_URL (required)

  • ALLURE_TOKEN (required)

  • ALLURE_PROJECT_ID (optional. default project = 37)

  1. Build the project:

npm run build
  1. Start the MCP server from the compiled output:

npm start
  1. Optional: run in development mode (TypeScript via tsx):

npm run dev
  1. Optional: run integration smoke checks:

npm run test:integration

MCP Client Setup

Use one of these server commands:

Common config block:

{
  "mcpServers": {
    "allure-testops": {
      "command": "npx",
      "args": ["-y", "github:{repo-name}/allure-testops-mcp"],
      "env": {
        "ALLURE_TESTOPS_URL": "https://{allure-testops-instance-address}",
        "ALLURE_TOKEN": "{your-api-token}",
        "ALLURE_PROJECT_ID": "{allure-project-id}"
      }
    }
  }
}
  • command: node

  • args: ["/absolute/path/to/allure-testops-mcp/dist/index.js"]

  • example: node c:\users\username\allure-testops-mcp/dist/index.js

  • command: npx

  • args: ["-y", "github:iampopovich/allure-testops-mcp"]

  • Open Claude Desktop MCP settings and add the mcpServers JSON entry above.

  • Restart Claude Desktop after saving config.

  • Add the same mcpServers entry in your Claude Code MCP configuration.

  • Restart your Claude Code session to load the server.

  • Open Cursor MCP settings and add the same mcpServers entry.

  • Restart Cursor (or reload MCP servers) after saving.

Any MCP client that supports stdio servers can use this project with the same command/env configuration.

CI and Quality Gates

This repository includes GitHub Actions checks for pushed code and pull requests:

  • compile (npm run build)

  • lint (npm run lint)

CI workflow: .github/workflows/ci.yml

OSS Automation

Included automation for typical open-source maintenance:

Integration Smoke Test

ALLURE_TESTOPS_URL="https://allure-testops.instance.com/" \
ALLURE_TOKEN="your-api-token" \
ALLURE_PROJECT_ID="37" \
npm run test:integration

Public Project Docs

Available Tools

100 tools
add_test_cases_to_launchC

Add test cases to a launch.

Inputs: id (required), payload (required).

add_test_case_tags_bulkC

Add one or multiple tags to one or multiple test cases using bulk API.

Inputs: projectId, projectName, testCaseId, testCaseIds, +2 more.

add_test_plan_to_launchC

Add a test plan to a launch.

Inputs: id (required), payload (required).

apply_defect_matchersA

Apply all configured defect matcher rules to a launch. This auto-triages unresolved test results by matching them to existing defect records based on error message patterns. Run this after a launch completes to auto-group failures.

Inputs: id (required).

archive_shared_stepA

Archive a shared step to retire it from active use. Archived steps remain readable but are hidden from the active library. Run get_shared_step_usage first to check impact before archiving.

Inputs: id (required).

assign_test_resultC

Assign a test result. payload must include username.

Inputs: id (required), payload (required).

bulk_close_defectsA

Close multiple defects at once — mark them as resolved after a fix is confirmed. payload must include: ids (array of defect IDs).

Inputs: payload (required).

bulk_reopen_defectsA

Reopen multiple closed defects — use when a regression is detected after a fix was marked complete. payload must include: ids (array of defect IDs).

Inputs: payload (required).

close_launchB

Close an open launch.

Inputs: id (required).

copy_dashboardB

Copy an existing dashboard to the same or another project.

Inputs: id (required), name, projectId, projectName, +1 more.

create_dashboardB

Create a new dashboard in a project.

Inputs: name (required), projectId, projectName, shared, +1 more.

create_defectA

Create a new defect record to group a novel failure pattern. payload must include: name (string), projectId (number). Optional: description (string).

Inputs: payload (required).

create_launchB

Create a new launch. payload.projectId defaults to ALLURE_PROJECT_ID env when omitted.

Inputs: payload (required).

create_shared_stepA

Create a new shared step in a project. payload must include at minimum: name (string) and projectId (number). Use this to extract repeated test steps into a reusable library.

Inputs: payload (required).

create_test_caseA

Create a new test case. payload.projectId defaults to ALLURE_PROJECT_ID env when omitted. payload.precondition (string) is the preconditions field — always use this for preconditions/prerequisites text, NOT payload.description. payload.description (string) is the general test case description — always populate this with a meaningful summary of what the test case verifies. payload.customFields supports values like { customField: { id }, id, name }. payload.steps is an optional array of step objects with { name (step text), expectedResult (optional expected result text) } — steps are created after the test case. When a step has multiple expected results, list them in expectedResult separated by semicolons (e.g. "Result A; Result B") or as a bullet list (lines starting with -, *, or a number).

Inputs: payload (required).

create_test_planC

Create a new test plan. payload.projectId defaults to ALLURE_PROJECT_ID env when omitted.

Inputs: payload (required).

create_test_resultC

Create a new test result.

Inputs: payload (required).

delete_dashboardB

Delete a dashboard by ID.

Inputs: id (required).

delete_launchB

Delete a launch by ID.

Inputs: id (required).

delete_test_caseB

Delete a test case by ID.

Inputs: id (required).

delete_test_case_attachmentB

Delete an attachment from a test case by attachment ID.

Inputs: attachmentId (required).

delete_test_planB

Delete a test plan by ID.

Inputs: id (required).

find_similar_failuresA

Find test results that match a defect pattern within a launch. Use this to detect flaky tests and group failures by root cause: pass launchId to scope the search, optionally pass a defectId to check match against a specific defect. Returns test results that are candidates for linking to the defect.

Inputs: launchId, defectId, page, size.

get_automation_chartB

Get automation trend chart data for a project (test automation coverage over time).

Inputs: projectId, projectName, tcRql, launchRql, +4 more.

get_dashboardB

Get a dashboard by ID.

Inputs: id (required).

get_defectA

Get a defect by ID. Returns the defect name, status, description, and matcher configuration.

Inputs: id (required).

get_defect_launchesA

Get launches in which a specific defect has appeared. Use this to understand the blast radius: is this defect appearing in every run or only in specific environments?

Inputs: id (required), page, size.

get_defect_test_casesB

Get the test cases affected by a specific defect. Use this to understand which test scenarios are impacted by the failure pattern.

Inputs: id (required), page, size.

get_defect_test_resultsA

Get the test results grouped under a specific defect. Use this to see how many failures belong to this defect and in which launches they appeared.

Inputs: id (required), page, size.

get_group_by_automationB

Get test case counts grouped by automation status for a project.

Inputs: projectId, projectName, tcRql.

get_group_by_statusB

Get test case counts grouped by status for a project.

Inputs: projectId, projectName, tcRql.

get_launchB

Get a launch by ID.

Inputs: id (required).

get_launch_defectsA

Get all defects present in a specific launch. This is the fastest way to get a defect summary for a launch: 'what are all the distinct failure patterns in this test run?'

Inputs: id (required), page, size.

get_launch_duration_histogramB

Get histogram of launch durations for a project.

Inputs: projectId, projectName, tcRql, launchRql, +3 more.

get_launch_progressC

Get launch progress widget data.

Inputs: id (required).

get_launch_statisticC

Get launch statistics.

Inputs: id (required).

get_mute_trendB

Get trend of muted test cases over time for a project.

Inputs: projectId, projectName, from, to, +1 more.

get_shared_stepA

Get a shared step by ID. Returns metadata: name, project, archived status.

Inputs: id (required).

get_shared_step_stepsA

Get the full normalized scenario (list of steps) inside a shared step. This is the key tool for resolving sharedStepId references in test case scenarios: when get_test_case_steps returns a step with sharedStepId, call this tool to inline the actual step content. Returns the same NormalizedScenarioDto format as get_test_case_steps.

Inputs: id (required).

get_shared_step_usageA

Get the list of test cases that use a specific shared step. Use this for impact analysis before editing or archiving a shared step: shows which test cases would be affected.

Inputs: id (required), page, size.

get_statistic_trendC

Get test result statistic trend over time for a project.

Inputs: projectId, projectName, tcRql, launchRql, +4 more.

get_tc_last_resultB

Get last test result for each test case in a project.

Inputs: projectId, projectName.

get_tc_success_rateB

Get test case success rate analytics over time for a project.

Inputs: projectId, projectName, tcRql, launchRql, +4 more.

get_test_caseB

Get a test case by ID.

Inputs: id (required).

get_test_case_attachment_contentA

Download the binary content of a test case attachment. Returns base64-encoded content with its MIME type.

Inputs: attachmentId (required).

get_test_case_custom_fieldsC

Get custom field values for a test case.

Inputs: testCaseId (required), projectId, projectName.

get_test_case_historyC

Get test case run history.

Inputs: id (required), page, size, sort.

get_test_case_issuesB

Get linked issues for a test case.

Inputs: testCaseId (required).

get_test_case_overviewC

Get test case overview data.

Inputs: testCaseId (required).

get_test_case_scenarioC

Get scenario for a test case.

Inputs: id (required).

get_test_case_stepsA

Get manual scenario steps for a test case. Returns a normalized scenario with root step and a flat map of all steps (scenarioSteps), where each step contains body, expectedResult, children IDs, and optional sharedStepId.

Inputs: id (required).

get_test_case_tagsB

Get tags assigned to a test case.

Inputs: testCaseId (required).

get_test_planA

Get a test plan by ID.

Inputs: id (required).

get_test_resultB

Get a test result by ID.

Inputs: id (required).

get_test_result_attachment_contentA

Download the binary content of a test result attachment. Returns base64-encoded content with its MIME type. Use attachment ID from list_test_result_attachments. For text-based attachments (logs, JSON, XML): decode base64 to read the content. For images (screenshots): the base64 PNG/JPEG can be rendered directly.

Inputs: attachmentId (required).

get_test_result_env_varsA

Get the environment variable values recorded for a specific test result. Shows what configuration context (browser, OS, env name, build) this result ran against. Use this when investigating a failure to understand if the environment is a factor: e.g. 'does this only fail on Chrome 124 but not on Firefox?'

Inputs: testResultId (required).

get_test_result_historyC

Get history for a test result.

Inputs: id (required), page, size, sort.

get_test_result_retriesA

Get the retry history for a test result. Returns the list of previous attempts for the same test in the same launch. Key tool for flaky test detection: if a test failed on attempt 1 but passed on attempt 2, it is a flaky test, not a real failure. Check status across retries to assess reliability.

Inputs: id (required), page, size.

get_widget_dataC

Get data for a specific dashboard widget.

Inputs: id (required), projectId, projectName, from, +1 more.

list_custom_field_valuesC

List values for a custom field in a project.

Inputs: customFieldId (required), projectId, projectName, query, +5 more.

list_dashboardsC

List all dashboards for a project.

Inputs: projectId, projectName, page, size.

list_defectsA

List defect records for a project. Defects group similar test result failures by root cause. Each defect has a name, status (open/closed), and optional matcher rules. Use this as the entry point for failure pattern analysis: 'what are the top open defects in this project right now?'

Inputs: projectId, projectName, status, name, +2 more.

list_env_varsA

List all environment variable definitions (keys) available in the system. Environment variables track test configuration context: browser, OS, environment name, build number, etc. Use this to discover what ev[] keys are available before constructing AQL filters like ev["browser"] = "Chrome" or ev["os"] = "Linux" in search_test_results or search_launches.

Inputs: no named parameters.

list_env_var_schemasA

List environment variable schema definitions for a project. Schemas define which env var keys are tracked and displayed for launches in this project. Use this to understand what configuration context is captured per launch in a given project.

Inputs: projectId, projectName, page, size.

list_env_var_valuesA

List all recorded values for a specific environment variable key. Returns the distinct values that have been used across test runs (e.g. all browser versions recorded). Use envVarId from list_env_vars. Useful for building precise AQL filters: know that ev["browser"] has values ["Chrome 123", "Firefox 115"].

Inputs: envVarId (required).

list_launchesC

List launches for a project.

Inputs: projectId, projectName, search, filterId, +3 more.

list_project_custom_fieldsB

List custom fields configured for a project.

Inputs: projectId, projectName, query, page, +2 more.

list_shared_stepsA

List shared steps (reusable step library) for a project. Shared steps are referenced from test case scenarios via sharedStepId. Use this to discover the available shared step library before reading test case steps.

Inputs: projectId, projectName, search, archived, +2 more.

list_test_case_attachmentsB

List attachments for a test case.

Inputs: testCaseId (required).

list_test_casesC

List test cases for a project.

Inputs: projectId, projectName, search, filterId, +3 more.

list_test_plansC

List test plans for a project.

Inputs: projectId, projectName, search, page, +2 more.

list_test_result_attachmentsA

List attachments (screenshots, logs, HAR files, etc.) for a test result. Returns attachment metadata: id, name, contentType, size. Use this to discover what evidence is available before calling get_test_result_attachment_content.

Inputs: testResultId (required), page, size.

list_test_resultsC

List test results for a launch.

Inputs: launchId (required), search, filterId, page, +2 more.

remove_test_case_tags_bulkB

Remove one or multiple tags from one or multiple test cases using bulk API.

Inputs: projectId, projectName, testCaseId, testCaseIds, +2 more.

reopen_launchC

Reopen a closed launch.

Inputs: id (required).

resolve_test_resultB

Resolve a test result. payload must include status.

Inputs: id (required), payload (required).

restore_test_caseA

Restore a deleted test case.

Inputs: id (required).

run_test_planC

Run a test plan by ID.

Inputs: id (required), payload.

search_launchesB

Search launches by AQL query.

Inputs: rql (required), projectId, projectName, page, +2 more.

search_test_casesB

Search test cases by AQL query.

Inputs: rql (required), projectId, projectName, page, +2 more.

search_test_resultsA

Search test results by AQL query. The 'id' field in returned results is an integer — pass it as a number (not a string) to get_test_result or get_test_result_retries.

Inputs: rql (required), projectId, projectName, page, +2 more.

set_test_case_custom_fieldsB

Add custom field values for a test case via bulk API. Supports grouped values [{ customField: { id }, values: [{ id|name }] }] and flat values [{ id|name, customField: { id } }].

Inputs: testCaseId (required), payload (required), projectId, projectName.

set_test_case_issuesC

Set linked issues for a test case.

Inputs: testCaseId (required), payload (required).

set_test_case_tagsC

Set tags for a test case.

Inputs: testCaseId (required), payload (required).

suggest_env_varsA

Search environment variable definitions by name. Returns matching env var keys with their IDs. Use this to find the exact name of an env var key before querying values.

Inputs: query, page, size.

suggest_env_var_valuesA

Search for recorded environment variable values by partial text. Optionally scope by envVarId, projectId, or launchId. Use this to autocomplete valid values for AQL ev[] filters before running a search.

Inputs: query, envVarId, projectId, projectName, +3 more.

unarchive_shared_stepB

Restore an archived shared step back to active status.

Inputs: id (required).

update_dashboardB

Update dashboard fields (name, shared status, etc.).

Inputs: id (required), payload (required).

update_defectB

Update defect name, description, or status.

Inputs: id (required), payload (required).

update_launchC

Update an existing launch.

Inputs: id (required), payload (required).

update_shared_stepC

Update shared step metadata (name, etc.) by ID.

Inputs: id (required), payload (required).

update_test_caseB

Update an existing test case. payload.customFields supports values like { customField: { id }, id, name }.

Inputs: id (required), payload (required).

update_test_case_stepB

Update a test case step by ID. Supports updating body text and/or expectedResult. If expectedResult is provided, the expected result node is created automatically if it does not exist yet.

Inputs: stepId (required), body, expectedResult.

update_test_planC

Update an existing test plan.

Inputs: id (required), payload (required).

update_test_resultC

Update an existing test result.

Inputs: id (required), payload (required).

upload_test_case_attachmentA

Upload a file attachment to a test case. Provide file content as a base64-encoded string.

Inputs: testCaseId (required), filename (required), contentType (required), contentBase64 (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    C2.8/5.0

    Scored across 100 tools

    Disambiguation2/5

    The tool set spans many resource areas with clear individual descriptions, but multiple tools occupy close conceptual territory—get_test_case_scenario vs get_test_case_steps, get_launch_statistic vs get_launch_progress, and several trend/group analytics—making misselection likely without deep domain knowledge. The sheer volume of 100 tools increases the chance an agent picks the wrong one for a task.

    Naming Consistency4/5

    All tools use snake_case and mostly follow the verb_noun pattern (get_, list_, create_, update_, delete_, set_, etc.). Minor inconsistencies like 'get_tc_last_result' vs 'get_test_case_history' and 'bulk_close_defects' vs 'add_test_case_tags_bulk' (prefix vs suffix placement) prevent a perfect score.

    Tool Count1/5

    100 tools for a single MCP server is an extreme number, far exceeding the 50+ threshold for unreasonable scope. Even though Allure TestOps is a feature-rich platform, this volume is unwieldy and would overwhelm agents during tool selection.

    Completeness4/5

    The surface provides broad CRUD and lifecycle coverage for test cases, launches, test results, defects, test plans, dashboards, and shared steps, plus analytics, bulk operations, and env-var tooling. Minor gaps exist (e.g., no delete test result, no step-level editing for shared steps), but they are workaroundable rather than dead ends.

    Maintenance

    ActivityMaintained
    ResponsivenessNo recent community issues

    Related MCP Connectors

    Related MCP Servers