Sparq:manual To E2e

Manual to E2E — Convert Manual Tests to Automated E2E Tests

Config, version check, pattern rules, and E2E code generation preamble per claude/rules/skills.md.

Workflow

1. Parse manual test cases from input:
   • Sparq markdown: path contains .sparq/test-cases/
   • TestRail XML: <sections> root element
   • Qase JSON: {"cases":[...]} or .sparq/tms-export/TC-{feature}-qase.json
   • Local JSON: .sparq/tms-export/ directory with test-cases.json
   • TestRail CSV: headers include Title, Steps, Expected Result (exported from TestRail > Export > CSV)
   • Inline text: numbered steps with expected results
   • TestRail MCP: user provides TestRail project/suite reference. Fetch via mcp__testrail__get_sections + mcp__testrail__get_cases. Requires inputs.tms.testrail or outputs.tms.testrail config.
   • Qase MCP: user provides Qase project/suite reference. Fetch via mcp__qase__list_suites + mcp__qase__list_cases. Requires inputs.tms.qase or outputs.tms.qase config.
   • Extract: test ID, title, preconditions, steps, expected results 1.5. TMS Read (when input is MCP-based):
   • Resolve provider: user-specified > inputs.tms.provider > outputs.tms.provider (fallback)
   • Normalize to SparQ format, write .sparq/test-cases/TC-{feature}-tms-import.md
   • Cases with no steps (custom_steps_separated: null or steps: []): set automationStatus: "not_automatable", include in markdown but skip E2E generation
   • If MCP fails: prompt for file export per degradation-strategy.md
2. Scan E2E directories (per e2e.structure.* from config) for existing page objects, helpers, fixtures, auth setup, naming conventions. Reuse -- never duplicate infrastructure.
3. If Figma MCP available and enabled: match UI elements to Figma components, extract data-testid values, map layers to DOM structure. If Figma MCP is unavailable, selectors are derived from codebase grep for data-testid attributes per degradation-strategy.md.
4. CHECKPOINT -- Clarify ambiguities: test data requirements, auth needs, unresolved UI elements, API mocking vs real backend. Wait for answers.
5. CHECKPOINT -- Delegate to sparq-automation-engineer agent with tech context: project.componentFileExtensions, e2e, project.sourceRoot, project.routeDiscoveryPattern, preferences.locatorPriority. Present generated code. Wait for approval. 5b. Source preservation: The manual-to-e2e conversion does NOT delete or modify source manual test files. If the user requests cleanup of source files after conversion, present a removal checkpoint per tms-abstraction.md <removal_policy> before any deletion.
6. Optionally verify with Playwright CLI (see "Browser Verification (Playwright CLI)" below)
7. Optional lint check: After files are written and smoke verify passes, offer sparq lint {e2e-directory}/ to validate converted tests against 8 deterministic code-quality rubrics (locator quality, flaky patterns, assertion coverage, naming conventions). Instant, CI-compatible, zero model inference.

Delegation: sparq-automation-engineer agent (with full tech context from config)

Browser Verification (Playwright CLI)

When e2e.framework is playwright: navigate to target URL, take accessibility snapshot and screenshot for selector and visual confirmation (e.g., npx playwright screenshot <url> --output=<path>). See playwright-cli-tools.md for tools. If unavailable, skip and note: "[sparq] Browser verification skipped -- Playwright CLI unavailable or E2E framework is not Playwright".

Output

{e2e.structure.pages}/{Feature}Page.ts        # Page Object Models
{e2e.structure.steps}/{feature}Steps.ts        # Reusable step helpers
{e2e.structure.specs}/{feature}.spec.ts        # E2E test specs
.sparq/coverage/coverage-matrix.md             # Coverage tracking (metadata)
.sparq/tracking/test-registry.json             # Test registry (metadata)

Files are written directly to the project E2E directory per e2e.structure.* config. Existing files are edited in-place. Git is the safety net for review (git diff) and revert (git checkout).

Generated specs follow project conventions from step 2. If no existing patterns, use standard conventions for the configured e2e.framework (e.g., test.describe/test with @playwright/test for Playwright, describe/it with Cypress). For Cypress: import commands from project support file, use cy.session() for auth, and follow cypress-patterns.md conventions.

References

• .claude/skills/sparq-shared/references/playwright-patterns.md, cypress-patterns.md, e2e-common-patterns.md
• .claude/skills/sparq-shared/references/data-model.md
• .claude/skills/sparq-shared/references/test-generation-patterns.md
• .claude/skills/sparq-shared/references/pattern-adherence.md
• .claude/skills/sparq-shared/references/degradation-strategy.md
• .claude/skills/sparq-shared/references/playwright-cli-tools.md
• .claude/skills/sparq-shared/references/testrail-formats.md
• .claude/skills/sparq-shared/references/qase-formats.md
• .claude/skills/sparq-shared/references/tms-abstraction.md
• .claude/skills/sparq-shared/references/config-schema.md
• .claude/skills/sparq-shared/references/error-handling.md
• claude/skills/sparq-qase-api/SKILL.md
• claude/skills/sparq-testrail-api/SKILL.md

Usage

/sparq:manual-to-e2e

Examples:

• "Convert manual tests at .sparq/test-cases/TC-login.md to E2E"
• "Automate my manual test cases from TestRail"
• "Turn these manual tests into Playwright specs"

Example

/sparq:manual-to-e2e .sparq/test-cases/TC-login-manual.md
-> parses 26 cases, scans e2e/ (3 page objects, auth fixture, 2 helpers)
-> enriches selectors from Figma
-> CHECKPOINT: 2 clarifications needed (admin role setup, edge case data)
-> generates LoginPage.ts + login.spec.ts (18 automated, 8 skipped manual-only)
-> writes files directly to project E2E directory per e2e.structure.* config
-> verifies selectors via Playwright CLI screenshot + accessibility snapshot (when e2e.framework is playwright)
-> output: e2e/pages/LoginPage.ts, e2e/specs/login.spec.ts

/sparq:manual-to-e2e --from testrail --project 1 --suite 42
-> fetches 8 cases from TestRail via MCP, normalizes to SparQ format
-> generates AuthPage.ts + auth.spec.ts (7 automated, 1 skipped: no steps)

/sparq:manual-to-e2e --from qase --project PROJ
-> fetches 6 cases from Qase via MCP, normalizes to SparQ format
-> generates AuthPage.ts + auth.spec.ts (5 automated, 1 skipped: no steps)