v1.1.0 · Released October 13, 2026
Requirements and traceability, Impact Analysis from your git diff, automation launched from the test run, and reports you can freeze.
A tree of what the system should do — synced or authored
Every project gets an opt-in Requirements area. Author requirements directly, or pick which tracker issue types count and TestPlanIt mirrors them — hierarchy included — from Jira, Azure DevOps, GitLab, GitHub, Gitea, Redmine, or MantisBT. Provenance badges, versioned content with per-field diffs, drag-and-drop reorganizing, and case links from either side.
Rollups, suspect flags, and snapshots for sign-off
Each requirement rolls up the latest result of every case covering it — scoped to a milestone or configuration when you need it — and a link is flagged Suspect when the requirement changes after its case last ran. Traceability, Coverage Gaps, and Coverage Changes reports put it in front of stakeholders, and a Snapshot freezes the matrix so sign-off has evidence, not a screenshot.
Test what changed, not everything
Connect the repository of the application under test — GitHub, GitLab, Bitbucket, Azure DevOps, or Gitea — pick two commits or a pull request, and get the Affected Tests where you compose the run. It works from the git diff alone: no coverage agent, no instrumentation. Code Pins, ticket keys in commits, history, and an AI ranking each explain why a test was selected, and repository webhooks can compose the run for you.
Run your automation from the test run
Click Execute automated cases and the automated part of a run executes in your CI — a GitHub Actions workflow, a GitLab CI pipeline, or a signed webhook for Jenkins, Buildkite, CircleCI, and others — while testers pick up the manual cases. Define the parameters your pipeline takes — a browser, an environment, whatever it needs — and choose their values when you execute. No agent to install, any test framework, and manual and automated results share one checklist.
Keep the report — and the numbers you signed off on
Save any report, pre-built or custom, and reopen it with every setting restored; relative ranges like Last week or Last N days keep their meaning each time it opens. Save or share a report frozen and every open shows exactly the numbers captured at that moment, stamped with when and by whom — including public and password-protected links.
Several identity providers on one instance
Let Okta manage employees and Entra manage contractors on the same instance: each directory owns its own users and groups, and a write across that line is refused and logged. Rotate SCIM tokens with an overlap window, map IdP-asserted roles to access tiers, and scope a group to the projects it works on.
Never lose a test case to a closed tab
Test case edits are kept as you type — in the Add Case dialog and in edit mode — and offered back the next time you open the case. Drafts stay private to you, and nothing reaches the published case until you click Save.
Map a CSV once, link tickets by key
Save a CSV import's column mapping — delimiter, encoding, row mode, and template included — and the wizard suggests it the next time a matching file arrives. A ticket key like PROJ-123 is now enough to link an issue from a CSV import, the API, or an AI agent on the MCP server, so a repository with thousands of ticket references migrates in one import.
64 new endpoints, and agents that keep your formatting
Everything new in 1.1 is reachable by API token — requirements, impact analyses, execution targets, and the automation plan a CI job fetches. Creates through the REST API, the MCP server, and imports share one validated writer with the web UI, rich text round-trips through the MCP server as Markdown, and results can be recorded against a single iteration of a data-driven case.
One more AI provider, one step to connect
DeepSeek gets its own integration with model selection, thinking mode, and cost defaults instead of the generic OpenAI-compatible workaround. Test generation got leaner too: shorter preconditions, one scenario per case, your named test data, and a retry for a single generated case.
Pick the issue types that count as requirements (Epic, Story, a custom type) per integration; GitHub and Gitea select repository labels instead. An impact preview shows what will change before you save, and removing a type is reversible.
Override the rule for a single issue: promote a bug into the requirements tree, exclude a story that shouldn’t be there, and reset either back to what its type says.
Synced epics, stories and sub-tasks keep their parent/child structure, completed within a single import run. Tracker priority is mirrored too — no more fabricated “medium”.
Release a synced requirement to local editing in one click. It keeps its tracker badge and link, and becomes as editable as a manual one.
Every change to a requirement’s title, description or documentation — edited here or arriving from a sync — is a numbered content version with per-field diffs, and snapshots record which version each requirement was at.
Rich-text notes and file attachments, staged with the rest of your edits and saved or discarded together — on synced requirements too.
Milestone and configuration pickers narrow the rollup to executions from matching runs — a release scope includes its sprints — in the list, the detail panel and both coverage reports.
The execution sheet shows the requirements a case covers, and a Linked Requirements panel on Test Case Details lets you link more from the case’s own project.
Above a threshold the tree loads a window at a time — roots first, children on expand, more rows as you scroll — with search and the Coverage, Status and Source filters resolved on the server and every match returned with its ancestor chain.
Grab handles, outlined drop zones and a root-level strip; invalid moves such as dropping a requirement into its own subtree are rejected server-side.
Deleting cascades through descendants and says exactly how many; it’s a soft delete, restorable from the Trash.
Requirement rows are kept out of issue lists, issue pickers, issue-tracking reports and the MCP issues tool by default — their coverage is reported by the requirement reports instead.
Pin a test case to a file, a line range, a function or class, or a glob pattern from the Code Pins panel on the case page. Any change inside a pinned block selects the case first; a pin whose lines no longer exist still matches at file level and is flagged Stale.
Put @testplanit case:123 in a comment above the code it covers, or list globs against case ids and tags in .testplanit/testmap.yml. Markers are scanned on every cache refresh and problems are reported on the settings page.
A commit naming PROJ-123, #42 or AB#42 selects every case linked to that ticket and leaves a Ticket pin on the functions it changed. Tickets TestPlanIt doesn’t hold yet are imported from the tracker, ready to link.
A ticket raised against a test result — or a single step result — can select that case as well, not just tickets linked to the case itself. It’s a switch on each repository connection’s Linked Tickets card, off by default.
Rescan Recent Commits and Scan Full History run on demand, report progress by stage, can be cancelled, and resume where a capped scan stopped — so pins backfill for tickets linked later.
Choose a pull request instead of two commits — filtered by Open, Merged or Closed — and it’s measured from the merge base, so the diff describes what the PR did and nothing that landed on the target branch meanwhile.
Connect a repository for every service the application spans, each with its own branch, path patterns, cache and pins. The Impact dialog, the Code Pins panel and the settings page gain a repository choice, and the repository page gets a Code Pins column and filter.
An opened pull request is analyzed against its target branch; a push to the base branch analyzes the pushed commits; optionally, a push to any branch is compared against base. Each analysis composes a run of the affected cases. GitHub, GitLab, Bitbucket, Azure DevOps and Gitea.
Each affected test carries a score and a Pinned, Affected or Related tier, with Pin, Ticket, Keyword, History and AI badges explaining why — including the AI’s rationale — so the team can trust the shortlist and add to it.
Start from a previous analysis to set the base to its head commit, paste any ref, or jump to the latest on a branch. The same two commits within 24 hours reuse the completed analysis.
The Affected Tests open at the selection threshold with a count in the header, and can be filtered, sorted and selected by tier, reason and score. Branch search reaches the provider directly, past the first 500 branches.
A background check reads each pinned file at the branch tip, and the settings page shows per repository how many Code Pins no longer match — with Remove Stale Pins to clean them up, sparing dismissed pins and those the repository’s own markers declare.
Project Settings → Automated Execution holds GitHub Actions workflow, GitLab CI pipeline and signed-webhook targets — Jenkins, Buildkite, CircleCI or a home-grown runner — with static inputs, a timeout and a Verify check that catches a bad credential or missing workflow input before you dispatch.
The job fetches GET /api/test-runs/{id}/automation-plan (or runs testplanit run plan) and gets every automated case with the id tokens, class and full name a runner filter can match on — so a run of any size fits, and nothing framework-specific lives in TestPlanIt.
A manual run becomes hybrid the first time automated results land in it. Each result updates its case, the progress bar and ready-to-complete check reflect both kinds of execution, and any case can still be executed by hand.
Execute automated test on a case page or the repository row menu creates a run holding just that case, which completes itself once the job finishes and its results are in.
An execution chip on the run shows Pending, Dispatched, Running, Succeeded, Failed, Could not start, Timed out or Cancelled, with a link to the job. Cancel asks the provider to stop; Retry reopens the dialog; Execution history survives run completion.
The Playwright and WebdriverIO reporters and the CLI treat a run named by TESTPLANIT_RUN_ID as externally managed: they attach results and never create or complete a run of their own. Two outbound webhook events and audit entries follow the lifecycle.
A target declares single-choice, multiple-choice, text or configuration parameters with defaults. The Execute dialog renders one control each, the choices reach the job as inputs, and they’re recorded on the execution chip, its history and the run’s documentation.
Select rows in a run and the button becomes Execute N selected automated cases; the dialog says how many selected manual cases it skips, and Retry replays the same subset.
Starting, retrying and cancelling an execution needs the new Automated Execution permission plus Test Run Results rights, so editing a run no longer means spending CI. Existing roles are backfilled from their results rights.
Create, read, update and delete execution targets with an API token, with the same validation, credential encryption and audit trail as the settings page.
A single dialog per linked project covers both a recent, capped sample and an uncapped All-history run of your requirement types — it tells you how many the tracker holds before writing anything, runs in the background, and can be stopped keeping everything imported so far.
A ticket key like PROJ-123 is all it takes to link an issue from a CSV import, the API or an AI agent on the MCP server — nobody has to open the ticket in TestPlanIt first, so a repository with thousands of ticket references migrates in one import.
Jira, Azure DevOps, Redmine and GitLab queries scope themselves to specific issue types, with GitLab paging each type independently; Azure DevOps and MantisBT imports page correctly past the first window.
Each synced issue keeps its own created date from the tracker, which powers the gap report’s Uncovered Since column.
Save a CSV import’s column mapping — with its delimiter, encoding, row mode and template — and apply it next time. The wizard suggests a saved mapping whose columns match, reports any it couldn’t apply, and shared mappings follow their template into every project. Test Cases, Shared Steps and datasets.
A new section on the active integration’s settings for the seven requirement-capable trackers; issue integration settings are merged into one card and Milestone Sync moves to an explicit Save.
A single Add inbound webhook flow covers issue-tracker and repository sources, both kinds list together with health, deliveries and Send test, and Bitbucket joins the inbound providers.
One row per requirement–case pair with parent path, coverage state, latest result and project, plus a donut of coverage states and a per-hierarchy coverage breakdown.
Zero-linked and never-run tiers, an On Closed Requirements count, debt by top-level requirement, and debt aging from under 30 days to over a year.
Pick a baseline snapshot and compare it to the live matrix or a later snapshot to see exactly which requirements’ coverage moved, and how.
Every gap row ends with a Generate Test Cases button that opens the AI wizard pre-seeded with that requirement — its title, body and, for synced requirements, the tracker description.
Administrators run the traceability and gaps reports across every project with requirements enabled at once, from Reporting & Analytics, with a Requirement Project filter in one shared menu.
Every analysis with its trigger, changed files, selection tiers, the run it composed and how that run went, plus weekly volume by trigger — per project or across projects.
Pins and uncovered changed files per repository directory, so you can see where the code is changing without a test pinned to it.
Trigger and Code Repository dimensions and three metrics — Impact Analysis Count, Affected Cases Selected and Selection Precision — with drill-down to the analyses behind each number.
Scope either report to selected subtrees, filter traceability by coverage state, and share links carry the scope — and every pre-built report’s parameters now survive the share redirect.
Pick Last week, This month or the new Last N days, weeks or months, and a live saved report or share link keeps that meaning: open it next Monday and it reports next Monday’s last week, in the timezone of whoever saved it, with the viewer stating the dates it currently covers. Choose Fixed dates when the dates matter more than the window.
Every pre-built report’s filters live in a single Filters menu, and each accepts more than one value. Flaky Tests gains folder, tag, milestone, configuration and custom-field filters; Test Case Health, Automation Trends, Impact Analysis History and Code Pin Coverage get their own. Existing share links, saved reports and API calls keep working.
Show total line draws a dashed sum across a multi-series chart plotted over Date or Milestone, and it rides saved reports and share links. Grouped table rows now total their children’s duration metrics instead of averaging averages, marked with Σ and a row count.
A case created through the REST API, the MCP server or an import now goes through the same writer as Add Case, in one transaction: template defaults applied, options matched by id or name, numbers held to their range, dates and links validated, rich text stored in the shapes the web UI stores — and required or restricted fields enforced rather than quietly accepted.
Steps, Text Long fields, session missions and notes, milestone notes and issue, result and review notes come back from the MCP server as Markdown instead of flattened text, and are parsed back into the same document on write — so an agent that reads a case, edits one step and writes it back no longer strips the formatting from the others.
Record a result against one row of a data-driven case, with attempts numbered per iteration; a case-level result on a parameterized case is refused instead of overwriting the rollup. Failed results can carry the issue a project requires, and the first result moves the run out of its initial state.
Runs created through the API or MCP server — and cases added to a run afterwards — generate the per-row iterations of parameterized cases just as the Add Test Run dialog does. A new MCP tool lists a run case’s iterations, with sensitive parameter values redacted.
Sessions written through the MCP server record a version on create, edit and complete the way the browser does, through a new versions endpoint open to API tokens; results added in bulk or over the API record the case version they ran against.
Project Settings → Access lists everyone with access, the role that governs them in this project, and the grant that decided it — a user permission, a group permission, the project default or system access. Administrators get a link straight to the project’s Users tab to change it.
Add/Edit on the Project Settings area makes a role’s holders project admins of every project where it is their effective role, however the role reaches them — the name “Project Admin” is no longer special. The same check now also covers project LLM integrations, JUnit suites, share links, repository connections and the audit log.
Issue Integration and Forecasting governed nothing and are gone from the role editor, so a role’s grid matches what it actually controls.
Force SSO used to hide the password form; the credentials provider and the signup API now refuse a password sign-in while any provider forces it. Signup resolves access level and role on the server, and user records, API-token ownership and sign-in account links are administrator-only through the model API.
Set SIGNIN_ADMIN_CONTACT=false and the “Contact the System Administrator” line leaves the sign-in page and the endpoint behind it stops looking the address up. It’s read at request time, so prebuilt images pick it up without a rebuild.
Test case edits are kept in the browser as you type and on the server a few seconds after you stop, in the Add Case dialog and in edit mode. Reopen the case and choose Restore or Discard; nothing reaches the case until you click Save.
DeepSeek V4 Flash, V4 Pro and the Flash Vision preview get their own integration with model selection, thinking mode and cost defaults — one step instead of the generic OpenAI-compatible workaround.
Generation asks for short preconditions, one scenario per case and steps that start in the application under test, and uses the test data you named instead of inventing accounts and IDs — so cases arrive near the length teams were trimming them to. Linked issues can be fed to the prompt, and the wizard says how many existing cases it found for context and warns when there are none.
When AI test generation returns a case that failed review, retry just that one from the review step instead of regenerating the batch.
Pre-release builds now publish themselves: every push to a release branch cuts the next beta as a GitHub pre-release, self-host server and worker images on a beta channel tag, and the Helm chart at that version — so you can try the next release before it ships.
Leave the access-key variables unset and storage uses the AWS credential chain — EC2 instance role, EKS IRSA, ECS task role or SSO. Static keys still work, so MinIO setups are unaffected.
Sixteen more kinds of soft-deleted records — comments, share links, execution targets, traceability snapshots, review requests and dataset rows among them — can be restored or purged from Admin Trash. The item types are derived from the schema, and a test fails the suite when a new model is missing from Trash.
Every table moves to TanStack Table v9. On a 250-row repository page the browser does about 38% less main-thread script work and sorts about 12% faster; column pinning, row selection and drag ordering carry over unchanged.
UPLOAD_MAX_MB now takes effect on the official images as well as source builds, so raising the attachment ceiling is a variable change rather than a rebuild.
Your milestone page is now a release cockpit
Milestones sync straight from Jira Fix Versions and Sprints — issue membership, webhooks, live updates — with burndown, a release-readiness rollup, a per-case traceability matrix in the PDF export, and an activity log. From the TestPlanIt panel on any Jira issue, generate test cases and QuickScript automation without leaving Jira.
Screenshots become test-generation context
Test-case generation now accepts images: screenshots pulled from linked Jira issues or Azure DevOps work items, rich-text documents with embedded images, or page captures from a URL crawl. A multimodal pipeline translates images per provider so it works with whichever LLM you've connected, and a cross-project LLM Usage report shows admins exactly where AI spend goes.
What happened, not just how many passed
Per-run execution metrics and duration, an execution timeline that shows real parallelization across workers, within-run flaky detection that understands retries, and result filters to slice big runs. Abandoned runs close themselves, linked cases flip to automated when results arrive, and AI agents can read it all through the MCP server.
Access control at the query layer, heavy lists ~80× faster
1.0 moves the entire data and access layer to ZenStack v3, enforcing permissions at the query layer with Prisma removed. Two-phase pagination took the heaviest lists from ~22 seconds to ~270 ms, project access resolves once per request, and runs, users, and audit logs window, prefetch, and stream on scroll.
Self-host on Kubernetes without a build step
Public self-host images on GHCR plus a single-tenant Helm chart for EKS, GKE, AKS, or k3s. Opt-in read-replica routing splits Postgres reads from writes entirely in-app, and /api/health reports event-loop lag so operators see trouble before users do.
Filter across every dimension, then save the view
A dockable, drag-resizable case-details panel, a sortable latest-result column, columns that remember their order and width, stacked filters across templates, states, tags, issues, and custom fields, saved views, and a searchable, virtualized folder tree that stays fast at any size.
Request in bulk, decide from one inbox
Request approvals in bulk, track pending-review badges wherever a run or session appears, and work a dedicated Reviews Inbox with full case details — including the requests you sent, with reminders. Approving applies the workflow transition automatically, and reviews cancel themselves when their subject is deleted or moved.
BANKING-TC-1234 says what it is at a glance
Optional project-prefixed identifiers like BANKING-TC-1234 tell readers which project and record type a link points to — cosmetic and additive, so every existing link and API call keeps working. TestPlanIt links pasted into Slack, Teams, or iMessage now unfurl into a card that says what they point to, with record details private unless you opt in.
Arabic, Czech, and a WCAG AA dark theme
Arabic — with the entire interface mirrored right-to-left — and Czech bring TestPlanIt to 17 languages, with community translations managed in the open. A new accessible Dark theme meets WCAG AA, enforced by a CI contrast gate, and switching themes no longer flashes the previous one.
Lease dataset rows so parallel runs never collide
Turn any shared dataset into a reservation pool. A CI job acquires the next free row, provisions its real fixture, and releases it when done — every lease carries a TTL, so a crashed job never pins a row forever. The locking model you know from Jenkins Lockable Resources or GitLab resource_group, for test data.
Two more issue trackers, natively linked
Link test results to issues in Redmine and MantisBT — now joining Jira, GitHub, GitLab, Gitea/Forgejo/Gogs, Azure DevOps, and Simple URL. Connect the tracker your team already uses and keep issues in sync with your runs.
Turn Jira tickets into test coverage without switching tools
Generate AI-powered test cases right inside the Jira issue panel. TestPlanIt reads the ticket's full context — description, comments, linked issues, and existing cases — streams suggestions for review, and saves them as linked, versioned test cases without leaving Jira.
Your identity provider, in charge of your entire user list
Pair SAML 2.0 single sign-on with SCIM 2.0 provisioning that creates, updates, and deactivates accounts straight from your identity provider — no manual admin. Out-of-the-box support for Okta, Microsoft Entra, and OneLogin, plus any SCIM 2.0-compliant provider.
Push Playwright results straight to your test cases
An official Playwright reporter that links results to cases by annotation, tag, or title, optionally auto-creates cases and folders from your test.describe structure, and uploads screenshots, videos, and traces — no launcher or worker coordination required.
Know which manual test to automate next
An AI-powered report that ranks your manual test cases by automation value, so your team always knows where converting to automation pays off most.
Sign-off, built into the workflow
Request reviews on test cases and runs, track approvals, and keep an auditable sign-off trail — without leaving TestPlanIt or bolting on a separate tool.
One case, many runs, real coverage
Drive a single test case from a table of input rows. Per-case and project-scoped shared datasets with version pinning bring true data-driven testing to your suite.
Talk to your test data
An official Model Context Protocol server lets Claude, Cursor, and other AI assistants securely read and update your test cases, runs, and results — bringing test management into your AI workflow.
Results you can defend
Server-enforced result governance: required fields, edit windows, and justified result flips keep your run history trustworthy and audit-ready.
Stream test data into your data lake
Export endpoints plus a full webhook event catalog stream your test data straight into the warehouse, so you can build analytics on top of it.
TestPlanIt now speaks 13 languages
Full localization across 13 languages brings TestPlanIt to teams everywhere — with community translations managed in the open.
Two-way sync with your issue tracker — and anywhere else
Bidirectional sync keeps issues and test results in step, while outbound webhook delivery pushes events to anywhere you can receive them.
Stronger controls for self-hosted teams
Configurable password policies and a round of security hardening give administrators tighter control over how their instance is accessed.
Keep your test suite lean as it grows
Find and resolve redundant test cases and repeated step sequences, merge duplicates, convert shared steps into reusable groups, and get early warnings when new cases overlap.
AI-powered tagging, zero busywork
AI suggests tags for test cases, runs, and sessions from their content and linked-issue context — review suggestions before applying and reuse existing tags when they fit.
Manual to automation in seconds
Turn manual test cases into executable automation code. Pick a framework and export ready-to-run scripts for Playwright, Cypress, Selenium, pytest, Jest, and more — real code, no vendor lock-in.