Coverage Discovery
Problem
Before you write new test cases for a work item, you want to know whether its requirements are already covered by something in the test library. This helps you avoid creating duplicates and cluttering the library with test cases that already exist. Without this skill, the only way to check is manually: you leave Rovo chat, open the test library, and search it yourself. The test case search skill runs the same scan from inside chat, so you get an answer without leaving the conversation.
Goal
Chain several of the Zephyr Agent’s skills together so Rovo does the searching for you: pull the issue’s own content, turn it into a wide spread of search terms, search the library with each one, pull full details on whatever comes back, and then judge relevance against the issue itself, excluding anything already linked.
Workflow overview
Fetch the work item's details. This gives Rovo the summary, description, and acceptance criteria to work from, and content to break into search terms.
Check existing test coverage for the work item. This gives Rovo the list of already-linked test cases, so it can exclude them from the final proposal.
Search the test case library using the extracted terms. Rovo runs one search per term to find as many candidates as possible.
Fetch full details for the matches. This retrieves the objective and test script for each candidate, which Rovo needs to accurately judge relevance.
Ask Rovo to evaluate and propose. Rovo compares the candidates against the work item's content and the existing links and returns only genuine matches, with reasoning.
Step-by-step prompts
Step 1: Fetch the work item and extract search terms
Prompt:
Fetch the details for DEMO-26. Break the summary, description, and acceptance criteria into individual nouns, verbs, and attributes that could work as test case search terms.
Expected output: the work item's summary, description, and acceptance criteria, plus a list of extracted terms, for example, login, email, invalid, session, and timeout.
Step 2: Check what is already linked
Prompt:
What test cases are already linked to DEMO-26?
Expected output: a list of linked test case keys. Keep this list available. You need it to exclude duplicates later.
Step 3: Search the test case library using the extracted terms
Prompt:
Use each of the terms from Step 1 as a separate search in the test case library.
Expected output: Rovo runs one search per term and returns a numbered list of matching test case keys and names for each term. Because matching is on name and key text only, broad, generic terms return more candidates than exact phrases copied from the work item.
Step 4: Fetch full details for the matches
Prompt:
Get the full details, including objective and test script, for the test cases you found.
Name specific keys only if you want to narrow the results to particular test cases. Otherwise, Rovo retrieves full details for every match. In both cases, the expected output is the objective, precondition, and script for each test case, which is what Rovo needs to judge fit rather than relying on the name.
Step 5: Ask Rovo to filter and justify
Prompt:
Compare these test cases against DEMO-26's summary, description, and acceptance criteria. Exclude TC-45 and TC-46, because they are already linked to DEMO-26. Propose only the test cases that are genuinely relevant, and explain why each one fits.
Expected output: a shortlist of proposed test cases, each with a short explanation of why it fits.
Combined prompt
To run the whole flow in one turn, use the following prompt:
Fetch the details for DEMO-26. Deconstruct the summary, description, and acceptance criteria into a list of every individual noun, verb, and attribute mentioned. Use each one as a separate, broad search term in the Zephyr test case library to maximize coverage. For all matches found, fetch their full test case details, including objectives and test scripts, to evaluate their actual relevance. Propose only the most relevant test cases that are not already linked to the work item, and explain why each one fits the criteria.
Notes and Caveats
This flow works only with the Zephyr Agent for Rovo, not with the generic Rovo chat. Make sure that the Zephyr Agent for Rovo is selected before you run these prompts. Zephyr's skills are otherwise unavailable.
Search matches only against test case name and key text, not step content, objectives, or descriptions. A relevant test case can be missed if its name does not match any extracted term.
Favor broad, generic terms over exact phrases from the work item. A name must literally contain the term to match, so single words surface more candidates than compound phrases.
Archived test cases are excluded from search results.
Breaking the work item into every noun, verb, and attribute can produce a long list of terms, and each term is a separate search. For a content-heavy work item, this results in many calls in a single turn. Trim the list to the terms most likely to appear in test case names to keep the number manageable.
If a large number of matches are returned, fetching full details for all of them at once can hit response-size limits. If this happens, ask for details on a smaller batch instead of all matches at once.
You can also stop this flow after Step 3. Ask Rovo to return only the candidate list of keys and names, without full details. Then fetch details and ask for a relevance judgment only for the test cases you want to examine.