Requirement Gaps and Use-Case Journeys
Every other requirements screen reports what Shift-Left Studio found. The Gaps & conflicts tab reports what it could not turn into a test, and why.
View as MarkdownApplies to: All editions (requirements review is basic on Free; full on Professional and Enterprise; workflow limits apply on Free — 50 workflows, 10 steps each) · Web app and Desktop app · A role that can edit the project
Overview
Every other requirements screen reports what Shift-Left Studio found. The Gaps & conflicts tab reports what it could not turn into a test, and why. Each untested requirement is listed under exactly one reason, written in plain words, so a long list becomes a handful of explanations — each with an obvious next step.
Some requirements are not single requests at all. A use case such as "the customer creates an order, adds an item, and pays" is a journey: several calls in a fixed order. When a use case records its calls in order, Studio builds a workflow test from it automatically as soon as every step has a test of its own. The Journeys panel, directly under the gaps list, shows which journeys are ready, which are blocked, and exactly which step is missing.
Key concepts
| Term | What it means |
|---|---|
| Gap | A requirement that produces no test, listed with the reason. |
| Use case with ordered steps | A requirement of type Use Case that records the calls it makes, in order (for example, POST /orders, then POST /orders/{id}/items, then POST /payments). Only these can become workflows. |
| Journey | The sequence of steps a use case describes. |
| Workflow test | A test built on the workflow canvas that runs the steps in order and passes values between them. |
| Blocked journey | A journey that cannot be built yet, with the reason per step. |
| "No test" step | A step whose endpoint exists but has no test yet. The test can be generated. |
| "No endpoint" step | A step calling an endpoint your API specification does not declare. There is nothing to test until the endpoint exists. |
Before you begin
- Requirements must have been extracted from at least one document. See Working with Many Requirement Documents.
- Your API specification should be imported (Setup → API Specs) so requirements can be tied to endpoints.
- Tests are generated only from approved requirements.
Step 1 — Open the gaps list
- Open your project's requirements and select the Gaps & conflicts tab.
- At the top, if your project has several documents, the Across your documents panel lists disagreements between documents. See Duplicates and Conflicts Across Requirement Documents.
- Below it, the list is grouped into sections. Click a section heading to expand it. Each section shows a count and the reason in its subtitle; each row shows the requirement's id, title, and the sentence it came from.
If nothing is missing, you see No gaps found — every extracted rule is scoped, consistent and testable.
Step 2 — Work through each reason
Each untested requirement is reported under the first reason that applies, in this order.
| Section | What it means | What to do |
|---|---|---|
| Your documents disagree — settle it above | Two documents give this rule different values. No test is generated until you decide. | Settle it in Across your documents at the top of the tab. |
| Rejected by you | You rejected the rule. | Nothing, unless the rejection was a mistake. |
| Set aside by the quality gate | The rule scored below the quality threshold. | Edit it in the review list, then approve it. |
| Needs clarification | The rule was flagged as needing clarification. | Clarify it with the document author, edit, and approve. |
| Awaiting your decision | Still waiting for approval. Generation reads only approved rules. | Approve or reject it in the review list. |
| Not tied to any endpoint | Studio could not match the rule to an endpoint, so there is nothing to test it against. Expanded by default. | Link it to its endpoints (Step 3). |
| Multi-step journeys — build them to cover these | A use case with ordered steps, every step already tested. The workflow can be built now. | Build it from the Journeys panel (Step 5). |
| Covered by a workflow | A workflow already walks through this use case. | Open the workflow to see the steps. |
| Journey blocked | A use case with ordered steps, but some steps have nothing testing them yet. | See the Blocked list in the Journeys panel (Step 6). |
| Prose only — no field, status or ordered steps | A story or capability described in prose, naming no field, status, condition, or ordered calls. Neither a single test nor a workflow can be built. | Add the steps it takes, or link it to the endpoints it applies to. |
| Nothing concrete to assert | The rule states no field, expected status, or condition that can be checked. | Edit the rule to state what must be true (a field, value, or status). |
The tab also lists:
- Your document contradicts itself — one document states the same fact twice with different values. Fix the document, or edit one of the two rules so they agree.
- Rules that disagree with your API spec — a rule and the specification disagree, so one of them is out of date. The rule is not approved automatically until this is settled.
- Ambiguities the AI could not resolve, Information missing from your document, and Assumptions the AI had to make — the AI's own notes from extraction, shown as it wrote them. Use them as questions for the document authors.
Note: A disagreement between documents appears only once, in Across your documents, rather than also as a contradiction inside one document.
Step 3 — Link a requirement to its endpoints
For rules under Not tied to any endpoint:
- Open the review list (Show requirements on a document's row, or Review all requirements on the batch bar).
- Find the requirement. Its Applies to endpoints: line is highlighted as unresolved.
- Expand the line and tick each endpoint the rule applies to.
- The change is saved immediately. Approved rules that are linked produce tests the next time you generate for those endpoints.
Choose All endpoints (cross-cutting rule) only for rules that apply everywhere, such as an authentication rule.
Tip: The Project Assistant can suggest the most likely endpoints for a requirement, with the evidence for each, and offer to link them for you to accept.
Step 4 — Fill coverage gaps automatically (optional)
On the Coverage tab, click Fill coverage gaps above the per-endpoint coverage matrix. Studio synthesizes requirements for uncovered areas of the project and reports how many it created, or that no new gaps could be closed automatically. Gap filling cannot help a requirement that is not tied to any endpoint; link those first. See Coverage and Extraction Fidelity.
Step 5 — Build workflow tests from ready journeys
The Journeys panel sits directly under the gaps list: "A journey is several calls in order. It becomes a workflow test once every one of its steps has a test of its own."
- Look at Ready to build (N). Each journey shows its title, id, and "N steps, every one of them already tested."
- Click Build N workflow tests.
- A confirmation reports "Built X and updated Y workflow test(s)." Any journey that could not be built is explained under Blocked.
- Open the new workflows from your project's workflows to see the steps on the canvas.
You often do not need to click at all. Every time tests are generated, Studio checks the project's journeys a few seconds later and builds any journey whose steps are now all tested. Running it again updates existing journey workflows rather than duplicating them.
Step 6 — Unblock a blocked journey
Each entry under Blocked (N) gives the exact reason. The most common reasons:
| Reason shown | Meaning | What to do |
|---|---|---|
| "Nothing tests POST /orders yet … Its endpoint exists, so the test can be generated." | A no test step. The endpoint is in your specification but has no test. | Generate tests for that endpoint. The journey builds itself shortly afterwards. |
| "The specification declares no POST /payments — the document this use case came from describes a call the API is not documented to have…" | A no endpoint step. The requirement names a call your API does not document. | Add the endpoint, import the specification that has it, or correct the requirement. |
| "It records no ordered steps, so there is no sequence to build." | The use case did not record its calls. | Rewrite the document so it states the calls in order, then extract again. |
| "It records one step, and a single call is a test rather than a journey." | Only one step. | Test it as a single request instead. |
| "Fewer than two of its steps name a path…" | The steps do not name endpoints. | State the method and path for each step in the document. |
If the project has no use cases with ordered steps, the panel shows No journeys in this project.
Below the Journeys panel, End-to-end flows this project could have suggests flows read from your specification and existing tests. Nothing there is created until you choose to build it.
Step 7 — Build a journey by name with the Project Assistant
You can ask the Project Assistant for a specific journey, for example "Build the workflow for REQ-4F21A9C0".
- If every step is tested, the assistant offers a card that builds that workflow. Accept it to build.
- If some steps have no test but their endpoints are in this project, the assistant offers one card that generates the missing tests and then builds the workflow. Accept it once; the workflow appears when generation finishes.
- If any step has no endpoint, the assistant explains which call the specification does not declare instead of offering to build.
The assistant never builds anything without your acceptance, and it will not generate tests into another project.
Understanding the results
| What you see | Meaning |
|---|---|
| A requirement under a gaps section | It produces no test, for the reason in the section's subtitle. |
| Journey under Ready to build | Every step is tested; building takes one click or happens automatically after the next generation. |
| Journey under Blocked | At least one step lacks a test or an endpoint; the reason names each one. |
| Requirement listed as covered by a workflow | A workflow for this journey exists. |
| A built journey that spans projects | One or more steps run a test from another project you have access to; the workflow spans both. |
Troubleshooting
| Symptom | Why it happens | What to do |
|---|---|---|
| Many user stories are listed under Prose only | They describe behavior without fields, statuses, or ordered calls. | Rewrite key stories as use cases with ordered calls, or link them to endpoints. |
| A journey stays blocked after generating tests | Another step is still missing a test or an endpoint. | Read the updated reason under Blocked. |
| The Build … workflow tests button is missing | No journey is ready. The button appears only when it would build something. | Resolve the blocked steps first. |
| A workflow's steps run but pass no values between them | No earlier step returns a value that a later step needs, according to the specification. | Open the workflow and use the workflow doctor to check its data connections. |
| A rule under Not tied to any endpoint is clearly about one endpoint | The rule does not name a field or operation the endpoint declares. | Link it by hand (Step 3). |
| A gap persists after you approved the rule | Tests have not been generated since approval. | Generate tests for the endpoint. |
Best practices
- Start with Not tied to any endpoint and Awaiting your decision; they usually account for most gaps.
- Write use cases with the method and path of each call, in order. That is what lets Studio build them as workflows.
- Treat no endpoint steps as questions for the API team: either the specification is incomplete or the requirement describes a call that does not exist.
- Share the AI's ambiguity and missing-information notes with document authors to improve the next version of the document.
- Revisit this tab after each extraction and each generation run.
Related articles
- Working with Many Requirement Documents
- Duplicates and Conflicts Across Requirement Documents
- Importing and Reviewing Requirements
- Coverage and Extraction Fidelity
- Requirements Traceability Matrix
For administrators (self-hosted installations)
| Setting | Default | What switching it does |
|---|---|---|
USE_CASE_JOURNEY_AUTOSYNC_ENABLED | true | false stops building journeys automatically after test generation. The Build … workflow tests button still works. |
Related articles
- Understanding Rule Intelligence · Product documentation
- Importing and Reviewing Requirements · Product documentation
- Test Data and Generation Setup · Product documentation
- Coverage and Extraction Fidelity · Product documentation
- Requirements Traceability Matrix · Product documentation
- Continuous Learning Insights · Product documentation
Next steps
- Getting started · Install + connect your spec
- Configuration fundamentals · Stabilize runs
- Initial configuration · Users, licensing, projects
- Release notes · Updates and fixes
Still stuck?
Tell us what you’re trying to accomplish and we’ll point you to the right setup—installation, auth, or CI/CD wiring.