Ln 522 Manual Tester

What it does

This skill simulates a manual tester (Ln 522) to verify acceptance criteria (AC) on running code and report structured results for quality gates. It creates executable test scripts, runs checks via bash/curl or Puppeteer, saves these scripts permanently for regression testing, and documents results in Linear. The skill focuses on mimicking human inspection processes within a defined workflow.

When to use it

• When a Story needs hands-on acceptance-criteria verification before automated planning.
• If the comment "## Test Research:" exists on a Story (as identified by ln-521).
• For all implementation tasks in a Story with a status of “Done”.

Key capabilities

• Creates executable test scripts and saves them permanently in tests/manual/.
• Runs AC-driven checks using bash/curl for APIs or Puppeteer for UIs.
• Documents results in Linear, including pass/fail status per AC and script path.
• Uses "expected-based testing" comparing actual results against reference files.
• Stores test results persistently in tests/manual/results/.

Example prompts

• “Test Story [Story ID]”
• “Verify acceptance criteria for Story [Story ID]”
• “Run manual tests on the code for Story [Story ID]”

Tips & gotchas

• Mandatory Read: The skill requires loading shared/references/ci_tool_detection.md to configure bash/curl/Puppeteer scripts correctly.
• Fail-Fast Principle: Tests must return a failure code (1) immediately if any criterion is not met, avoiding silent failures or warnings unless explicitly justified.
• Expected File Validation: Never blindly copy test results into the expected/ folder; always validate their correctness first.