testskill.config-example.toml

# Testing Skill Configuration — copy this file to testskill.config.toml in your project root
# Run /testskill.init to do this automatically.
#
# Keys come in three flavors:
#   1. Tool-agnostic defaults — already filled in with a sensible value, adjust only if needed.
#   2. Framework-dependent — left blank; pick the value matching your test framework (Playwright / Vitest browser mode).
#   3. Project-dependent — left blank; fill in with your project's commands, URLs and paths.

[workflow]
# When true, stops the workflow if no test plan (test-plan.md) is found before writing tests.
# This ensures tests are never written without a proper plan first.
stop_when_no_plan = true

# When true, stops the workflow if no app analysis (app-analysis.md) is found.
# App analysis captures runtime information like network calls that inform test writing.
stop_when_no_page_analysis = true

# When true, requires human review task and approval of test plans before proceeding to code tests
must_have_human_review_of_plan = false

# How many times to re-run tests to check for flakiness
how_many_executions_to_ensure_stability = 3

# Whether to run all tests for verification or only the subset of files related to the changes
run_only_specific_test_files = true

# When true, if page_analyzer_mode=runtime, stops the workflow if the app can not be started, executed and approached
# For example, when Playwright MCP is unable to access the target page, or the API client is unable to invoke a route
stop_when_cant_reach_runtime = true

# How many maximum tests can be planned and executed
# Unless dealing with small components/routes, it is highly recommended to keep this number low to avoid exhausting the context
max_amount_of_test_to_plan = 4

# How many maximum mutants (bugs) to embed in the code when verifying that tests fail when they should
number_of_mutants_per_test = 1

# Analysis mode for testskill.page-analyzer agent
# "runtime" = live browser inspection (default), "static" = source code only (fast), "auto" = runtime if no tests exist, static otherwise
page_analyzer_mode = "static"

# Thinking mode: controls depth vs speed tradeoff across all sub-agents
# "lean" = aim for good-enough results quickly, move forward fast
# "ultra" = thorough analysis, maximize coverage and confidence
thinking_mode = "lean"

[framework]
# Unit test framework for testing pure logic, utilities, hooks
# Vitest: "vitest"  |  Jest: "jest"
unit_test_framework = ""
# Vitest: "https://vitest.dev/guide/"  |  Jest: "https://jestjs.io/docs/getting-started"
unit_test_framework_docs = ""

# Library used for mocking functions, modules, and code-level dependencies
# Vitest: "vitest"  |  Jest: "jest"  |  Sinon: "sinon"
unit_test_mocking_framework = ""
# Vitest: "https://vitest.dev/guide/mocking.html"  |  Jest: "https://jestjs.io/docs/mock-functions"
unit_test_mocking_framework_docs = ""

# Framework for integration/page tests (browser-based testing)
# Playwright: "playwright"  |  Vitest browser mode: "vitest-browser-mode"
integration_test_framework = ""
# Playwright: "https://playwright.dev/docs/intro"  |  Vitest browser mode: "https://vitest.dev/guide/browser/"
integration_test_framework_docs = ""

# Library used for intercepting and mocking network requests
# Playwright: "playwright-route"  |  Vitest browser mode: "msw"
network_interception_framework = ""
# Playwright: "https://playwright.dev/docs/mock"  |  MSW (Vitest browser mode): "https://mswjs.io/docs/"
network_interception_framework_docs = ""

[commands]
# Command to start the application for testing/inspection (project-specific).
# The app must be running for UI tests and page analysis.
# e.g. "pnpm dev"  or  "NODE_OPTIONS='--max-old-space-size=8192' pnpm dev --mode=staging"
app_start_command = ""

# Command to stop the application (optional, project-specific).
# Leave empty if the app stops automatically or doesn't need explicit stopping.
# e.g. 'pids=$(lsof -ti tcp:5173); [ -n "$pids" ] && kill -9 $pids'
app_stop_command = ""

# Command to run the test suite.
# This is the primary command used after writing tests to verify they pass.
# Playwright: "pnpm test:page"  |  Vitest browser mode: "pnpm test:browser"
test_run_command = ""

# Command to run the test suite with coverage reporting (lcov, html, etc.).
# Optional; leave empty to skip coverage.
# Vitest browser mode: "pnpm test:browser:coverage"
test_run_command_with_coverage = ""

# Command to run a single test file
# Used when run_only_specific_test_files is true, also used for diagnosis and verification
# Playwright: "pnpm test:page {file name}"  |  Vitest browser mode: "pnpm test:browser {file name}"
test_run_single_file_command = ""

# Command to run a single test by file and title
# Playwright: 'pnpm test:page {file name} -g "{test title}"'
# Vitest browser mode: 'pnpm test:browser {file name} -t "{test title}"'
test_run_single_test_command = ""

# Flag to append to the test run command to capture a trace file
# Playwright: "--trace on"  |  Vitest browser mode: "--browser.trace on"
test_trace_flag = ""

# Flag to append to the test run command to repeat it multiple times for stability checks
# Playwright: "--repeat-each={repeat count}"
# Vitest browser mode: no built-in CLI flag; use a shell loop: "for i in $(seq {repeat count}); do <test command> || exit 1; done"
test_repeat_flag = ""

# Flag to disable test retries during stability checks — retries mask flaky tests by silently passing on a second attempt
# Playwright: "--retries=0"  |  Vitest: "--retry=0"
test_no_retry_flag = ""

# Command to stage changes for git.
git_stage_command = "git add -A"

[paths]
# The parent directory where to save all test sessions
# artifacts (test plans, app analysis, etc.)
# e.g. "./test/.context"
test_context_root_folder = ""

# Name of a session-specific context folder to save artifacts of a single test session. This folder will be saved underneath test_context_root_folder.
# This folder will contain: test-plan.md, app-analysis.md, test results, etc.
test_context_folder_name_template = "test-context-{feature-or-task-name}-{two digits suffix for uniqueness}"

[app]
# The base URL of the frontend application or API (project-specific)
# Used for navigation during analysis and testing
# e.g. "http://localhost:5173" (Vite default) or "http://localhost:3000"
app_url = ""
# Free-form instructions for authenticating during runtime analysis (project-specific)
# e.g. "Set localStorage item 'fake_store_is_authenticated' to 'true'"
login_instructions = ""

[outputs]
# Filenames for the output documents
# test-plan.md is generated by the testskill.planner agent
# page-analysis.md is generated by the testskill.page-analyzer agent
# test_result_report is generated by the testskill.verifier.md
test_plan_filename = "test-plan.md"
page_analysis_filename = "page-analysis.md"
test_result_report = "test-results.md"
html_report_filename = "test-results.html"
transcript_filename = "transcript.jsonl"

[patterns_and_practices]
# Maximum number of statements and expressions allowed in a single test
max_test_statements = 10

# Maximum number of assertions (expect statements) allowed in a single test
max_assertions = 3

# Maximum number of statements allowed in the Act phase of a test
max_act_phase_statements = 2

# Maximum number of lines for inline JSON/object literals in a test.
# Include only fields that are explicitly needed for the test's point. Everything else (unrelated defaults) should come from a data factory.
max_json_lines_in_test = 2

[flaky]
# How many times the test must pass consecutively to be considered stable
stability_pass_count = 10

# Max parallel workers to simulate saturation (escalation strategy)
saturated_workers = 2

# Artificial network delay in ms to add when simulating slow responses
artificial_delay_ms = 500

# CPU throttling rate for Chrome DevTools Protocol (1 = no throttle, 4 = 4x slower)
cpu_throttle_rate = 4

[canonical_example]
# A real test file that demonstrates all best practices.
# Test writers should read this before writing their first test.
# e.g. "src/pages/Cart/Cart.page.test.ts"
path = ""
# e.g. "When API returns a few products, then they are all visible in the cart"
title = ""