Existing coverage search
These three skills avoid redundant work: before reproducing a bug, repairing a selector, or generating new tests, they find out what's already covered by existing tests and which scenarios remain uncovered. They're read-and-reason only — none of them creates, modifies, or deletes files, or runs anything against the application.
lippia-neo-find-existing-tests
When to use it
Right after lippia-neo-read-user-story or lippia-neo-read-bug-report, within a bug cycle, to
search the tests directory for existing coverage of the related acceptance criterion or feature
— before reproducing the bug or generating anything new. It models the search in layers
(strong identity / controlled semantics), with an evidence strength per match
(strong/partial). It never aborts the cycle even if it finds nothing.
{
"input": {
"source_summary": {
"us_id": "PA-US-02"
}
},
"config": {
"tests_dir": "tests/",
"test_framework": "playwright"
}
}
| Input | Type | Description |
|---|---|---|
input.source_summary |
Required | Output of lippia-neo-read-user-story or of lippia-neo-read-bug-report. A bug cycle doesn't include acceptance_criteria/key_features. |
config.tests_dir |
Required | Directory where existing tests are searched. |
config.test_framework |
Required | Defines which globs and elements to inspect (playwright/lippia/pytest). With cypress it doesn't search yet: it returns not_found with a warning. |
input.issue_body |
Optional | Raw text of the bug, with the same alignment criterion as steps_to_reproduce/expected_result/actual_result. |
input.related_us_summary |
Optional | If source_summary.related_user_story is non-null and has already been resolved with lippia-neo-read-user-story; if it isn't provided, the search is degraded (this is recorded in warnings). |
Output
| Field | Description |
|---|---|
status |
found / not_found. |
existing_tests |
Confirmed matches (path, suite_type fe/be, matched_on, matched_location, match_strength strong/partial, match_mode exact/normalized, and test_case_id only if the file declares it); weak matches never go here. |
search_terms |
All the terms used, with term, source, priority, and match_mode, even if existing_tests ends up empty (traceability). |
warnings |
Degradations (bug without a story, unresolved related_user_story), Pytest fallback, discarded weak matches, nonexistent tests directory, etc. |
Chains with
Runs after lippia-neo-read-user-story/lippia-neo-read-bug-report; its existing_tests feeds
lippia-neo-generate-test-suite as a style reference, and lippia-neo-review-test-coverage to
determine what remains uncovered (qa-bug-coverage plan). In the qa-bug-fix-verification plan
it's used as config.target_tests of lippia-neo-run-and-heal, to run only those tests.
lippia-neo-find-ui-tests-by-code-diff
When to use it
To find out which existing tests match the part of the app that a git diff changed
(git diff in unified diff format of a PR against the target branch) — typically as step 1
of the selector repair flow, before deciding whether anything needs to be repaired. Unlike
lippia-neo-find-existing-tests (which starts from a user story), it derives its search
terms from the diff itself: touched files, added/removed/changed selectors, routes, and
CSS classes. A match doesn't mean "broken test", only that the test's scope overlaps with what
changed — it doesn't repair or run anything.
{
"input": {
"pr_diff_path": "tmp/pr-1234.diff"
},
"config": {
"tests_dir": "tests/",
"test_framework": "playwright"
}
}
| Input | Type | Description |
|---|---|---|
input.pr_diff_path |
Required | Path to a file with the git diff output of the PR vs. the target branch; primary source of the search terms. |
config.tests_dir |
Required | Directory where tests are searched. |
config.test_framework |
Required | Defines the globs and elements to inspect (playwright/lippia/pytest). With cypress it doesn't search yet: it returns not_found with a warning. |
Output
| Field | Description |
|---|---|
status |
found / not_found. |
has_matches |
Boolean equivalent to status == found, meant for a plan gate. |
matched_tests |
One object per file (path, suite_type always fe, test_case_id if the file declares it), with matched_on grouping ALL the terms that matched in that file (never just the most representative one); each term includes term, matched_location, match_strength strong/partial, and match_mode. |
search_terms |
Terms derived from the diff (changed_file, selector, route, css_class, text_content), with priority and match_mode (exact/normalized/template_regex; the latter is a pattern for texts with dynamic parts). |
warnings |
Empty/nonexistent diff, Pytest fallback, high-priority terms without a confirmed match, 100% interpolated text with no derivable pattern, etc. |
Chains with
Step 1 of the ui-selector-repair plan; its matched_tests is the mandatory input of
lippia-neo-repair-selectors-by-diff, which only repairs what appears in that list and only runs
if has_matches is true.
lippia-neo-review-test-coverage
When to use it
To determine which explicit scenarios of a user story or a bug remain without
coverage, based on a list of already-resolved matches (typically the output of
lippia-neo-find-existing-tests) and the declared acceptance criteria / key features.
It's pure reasoning over data already received: it doesn't search any tests directory or read
files.
{
"input": {
"source_summary": {
"us_id": "PA-US-02"
},
"existing_tests": [
{ "path": "tests/checkout.spec.ts", "match_strength": "strong" }
]
}
}
| Input | Type | Description |
|---|---|---|
input.source_summary |
Required | Same contract as in lippia-neo-find-existing-tests (story or bug); a bug with no associated story doesn't include AC/key_features. |
input.existing_tests |
Required | Array of already-resolved matches; only strong or partial matches count. |
input.related_us_summary |
Optional | AC/key_features/identifier of the associated story, if source_summary.related_user_story is non-null and was already resolved before invoking this skill. |
Output
| Field | Description |
|---|---|
needs_expansion |
true if any explicit scenario remains without strong/partial coverage, or if existing_tests came in empty. |
missing_cases |
Uncovered scenarios, each with description (as the story describes it, without rewording) and rationale (the exact phrase from the AC, key feature, or bug that backs it). For a bug with no associated story, the only scenario evaluated is the bug itself. |
warnings |
For example, related_user_story declared but not resolved. |
Chains with
Consumes the existing_tests of lippia-neo-find-existing-tests (qa-bug-coverage plan); its missing_cases feeds
lippia-neo-generate-test-suite, which switches to "scenario expansion" mode when that field is
present.