Release notes — August 2026
Our largest release to date — enterprise SSO and access control, data-driven testing across many rows of data, a rebuilt test-generation engine, the new Test Workbench, a CI/CD runner, and execution parity between the desktop Agent and the cloud Engine.
Summary
Build 2.0.96.222. Our largest release to date. Shift-Left Studio gains enterprise-grade access control and identity, data-driven testing that runs one test against many rows of data, a rebuilt test-generation engine that produces more meaningful tests, a new Test Workbench for editing and running tests side by side, and a CI/CD runner that can gate your build. Alongside the new capabilities, a large correctness pass makes the desktop Agent and the cloud Engine execute tests identically.
Improvements
Access control and identity
- Single sign-on: authenticate with OIDC, SAML, or Microsoft Entra ID, with just-in-time user provisioning and group mapping.
- Workspaces, per-project roles, and custom roles: grant granular permissions per project, and manage access by group.
- Platform API keys, outbound webhooks, and external secret managers: issue scoped keys, deliver webhooks with delivery history and redelivery, and pull credentials from an external secret manager.
- Comments on tests and endpoints: reviewers can leave feedback in place.
- Admin screens for SSO, API keys, webhooks, groups, and secret managers, each with a master on/off switch.
Data-driven testing
- Run one test against many rows of data: drive a single test or workflow from CSV, Excel, JSON, XML, YAML, databases, APIs, or generated sample data.
- Manage datasets in Project Settings, with preview, versioning, and import/export. Uploaded Excel test-data sheets become reusable datasets automatically.
- Data-driven runs work everywhere: on the desktop Agent, in workflows, and in scheduled packs, with per-iteration results and a "rows × dataset" summary in reports.
Test generation
- Rebuilt generation engine: a coverage-driven engine produces more coherent, behaviour-focused tests, including nested request-body fields.
- Outcome-based reporting: generation now reports what was created, updated, or skipped — and why — with attribution per testing technique, rather than exposing a pipeline.
- Requirements Traceability Matrix, per-endpoint gap attribution, and one-click gap fill.
- Better requirement extraction from business documents, including tables and narrative structure. Parsing now runs in the background so you can keep working.
- Cancel a test-generation run that is already in progress.
Authoring and execution
- New Test Workbench: edit, run, and inspect a test in one split view, or pop a test out into its own window.
- Assertion improvements: quick JSON assertions with a schema-aware field picker, dynamic values in expected results, a quality lint that flags unreliable assertions, and clearer failure explanations with a value-by-value diff.
- User-authored tags: tag tests, filter by tag, bulk-tag, and build run packs from tag rules.
- Cancel a running test pack, with live progress while it runs.
- Editable endpoint parameters, and support for the QUERY method on REST (off by default).
Pipelines and operations
- CI/CD runner: run a pack from your pipeline, emit JUnit results, and gate the build on the outcome.
- Operations: readiness and metrics endpoints, a downloadable support bundle, an on-prem backup script, and safer on-prem upgrades via schema migrations with a downgrade check.
- Optional anonymous product telemetry, off by default.
Navigation and performance
- Simple and Advanced modes, requirements as a first-class section, and deep links that reflect where you are.
- Faster reports and faster app load, plus a responsive layout for endpoint test pages on smaller screens.
- A problem in one section no longer blanks the entire app.
Fixes
- Execution parity: the desktop Agent and the cloud Engine now build the same request from the same test, field for field, and report results the same way. Previously some tests could pass on one and fail on the other.
- SOAP: tests now send exactly the XML they define, with custom headers, the correct content type for the SOAP version, faults reported as XML, and shared workbooks split per operation.
- GraphQL: local runs now use the GraphQL executor, and GraphQL configuration is saved correctly.
- Test generation correctness: removed unreachable "method not allowed" probes and assertions that could never pass, corrected wrong-data-type values that the API would have accepted, made coverage gaps closable, and made strategy and severity selections take effect exactly as chosen.
- Reporting: fixed drill-down history, print and preview, per-test scoping of data-driven results, reports mislabelled "Test Deleted", and run history appearing under the wrong test.
- Security: report endpoints now require authentication, activity streams are scoped to your own tenant, secrets are masked in logs and diagnostics, the server refuses to start with placeholder credentials, and development secrets are no longer included in the installer package.
- Installer and desktop app: the Windows installer no longer fails during code signing, the macOS installer is signed and notarised, the app no longer hangs when closing or reloading, and "Check for Updates" now reports a result instead of failing silently.
- Scheduling: editing part of a pack no longer clears its schedule, and scheduled reports in different formats now produce distinct emails.
- Sessions and navigation: expired sessions no longer trap you on a blank screen, you return to the page you intended after signing in, and search filters features, endpoints, and tests correctly.
- Imports: large API specifications import faster, test data can be imported from a file including Excel, and Postman and OpenAPI imports no longer create placeholder tests.
Known issues
- SOAP tests generated before this release may have stored the literal text
[object Object]as a request value. This release stops that from being generated, but existing tests keep the stored value until they are regenerated. Regenerate affected SOAP tests to pick up the correction. - The desktop Agent must be updated to this release. An earlier Agent will not include the execution-parity fixes above and can still produce results that differ from the cloud Engine.
Related links
Related articles
- Release notes — July 6, 2026 · Release notes
- Release notes — June 21, 2026 · Release notes
- Release notes — June 10, 2026 · Release notes
Helpful links
- All release notes · Browse by month
- Platform overview · Context for new users
- Getting started · Install + first run
- Configuration fundamentals · CI-ready stability
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.