Autonomous browser worker
Application discovery map
Configure a safe crawl scope, then let the agent build a living application map.
Ready
Ready for launchEstimated scope: 25 pages at depth 3.0%
Generate tests
Select Jira scope, enrich context, rate quality, and generate Playwright-ready test assets.
1ScopeChoose Jira work
2ContextAdd product detail
3RateReview quality
4Cases & scriptsGenerate outputs
Select Jira projects
ConnectedAvailable projects
Selections are combined into one generation run.
Generation scope
1 project selected
Scope summary
Your current selections and readiness for the next step.
Scope readiness25%
Projects selectedPortal
Jira scopeNot loaded
Generation modeEpics & stories
Confluence pages
spec contextAdd a Confluence page — all of its child pages are automatically fetched and included as context.
Figma files
design contextScreens and prototype flows will be used as additional context.
Existing Manual Test Cases · Optional
Excel / CSV coverageReuse existing coverage before generating anything new
Upload .xlsx, .xls, or .csv cases. SAGE validates, normalizes, maps them to requirements, and uses them to avoid duplicate tests.
No manual test file uploaded. This step is optional.
Default behavior preserves imported cases and generates only requirement coverage that is missing.
SAGE supplemental context
requirement packageThis context is attached to each selected requirement package and is included in DoR analysis and Azure OpenAI test generation. Jira remains the authoritative requirement source.
Building unified context and advisory DoR recommendations…
Calling Azure OpenAI — generating test cases…
Recorder
Record real browser sessions with Playwright, then replay & report — no code required.
SessionReady
Captured steps0
API calls0
Screenshots0
Elapsed00:00
Target Application
Linking a Jira issue turns the recording into lifecycle evidence. Without a Repository Test ID, SAGE creates a Draft candidate for review.
Recorder
Live CaptureIdle
Click Start Recording to begin capturing test steps
Actions, navigations, and screenshots will appear here in real time
API Calls by Page0 calls
API (fetch/XHR) calls will appear here grouped by page during recording
Loading test library...
Explorer agent
Enter any URL — the agent crawls same-origin pages, inventories elements & locators, captures API calls, and derives test cases.
SessionReady
Pages visited0
Elements found0
API calls0
Brittle locators0
Target Application
Discovered pages, APIs, locator risks, and test candidates are linked to these requirement packages.
Explorer Agent
Agent ActivityIdle
Enter a URL and click Start Exploring
The agent discovers pages, inventories elements, grades locators, captures API calls, and derives test cases automatically
Explorer Pro
Build a living application map from real browser behavior, UI semantics, network evidence, and testable user journeys.
Discovery evidence
Review live activity, page inventory, API usage, and candidate automation paths.
Figma design compare
Paste a Figma file/prototype link — every page & frame is inventoried for design elements, then matched against a real Explorer crawl to spot drift
Frames analyzed0
Design elements0
File—
Connect a Figma file
ReadyDesign analysis and drift evidence are linked to these lifecycle requirement packages.
Every canvas ("page") and every top-level frame ("screen") in the file is walked and inventoried for buttons, inputs, links and other interactive elements.
How comparison works
- 1Inventory design
Read frames, components, controls, and prototype paths.
- 2Match implementation
Compare the design model with the latest Explorer crawl.
- 3Review drift
Prioritize missing, changed, or unimplemented interactions.
AI Driver
Write tests in plain English — each step is resolved against the live app at runtime and self-heals when the UI changes. No selectors, ever.
Target Application
Test Steps (plain English, one per line)
Run History0 runs
No runs yet
Write some steps and click Run Test
Full testing lifecycle
Reconcile requirement intent, design, governed tests, runtime evidence, automation, execution, and release readiness in one traceable workflow.
Requirement packages
0 requirement packages
Loading lifecycle packages...
⌁
Select a requirement package
Review readiness, design, test coverage, runtime evidence, automation, executions, and open lifecycle gaps.
Test case repository
Persistent test library organized as Business Theme › User Story › Test Scenario. Classified generated cases are logged here automatically.
0%ACTIVE
Total cases
0
Active
0
Business themes
0
Test categories
0
Deprecated
0
Business themes
Test categories
Playwright script catalog
Generated Playwright specs saved for reuse — run them on demand and track the latest outcome.
Scripts cataloged0
Last run passed0
Last run failed0
Never run0
| Story | Business Theme | Category | Test Cases | Test Script | Pass/Fail | Last run date | Actions |
|---|---|---|---|---|---|---|---|
| Select this view to load the catalog. | |||||||
Triage a new story
Compare an incoming story against the repository — reuse what exists, update what drifted, retire what's obsolete.
Incoming story
from JiraStories & runs
Every Jira story and its test run status.
Total stories—
Done—
In progress—
| Story | Priority | Tests | Status |
|---|---|---|---|
| Select this view to load stories. | |||
Coverage
UI → API graph, tracked endpoints, and coverage gaps.
Total endpoints0
Covered0
Coverage gaps0
Coverage trend+4%
Coverage gaps
| Method | Endpoint | Reason | Impact | Last seen | Action |
|---|
Review queue
Low-confidence automatic classifications waiting on a human decision.
Medium impactStory analysis
PORT-2841 — reworded acceptance criteria
Behavior change or cosmetic? Expanded retest already ran (4/4 green).
High impactFigma drift
Primary CTA is 24px wider than design
Detected during the latest Figma comparison run on the statement details screen.
Low impactSelf-healed locator
Payee submit button locator auto-repaired
AI Driver replaced a stale selector with a role-based locator during the last run.
Medium impactGenerated cases
3 new negative scenarios need an owner
Generated from PORT-2855 but not yet assigned to a reviewer.
Approvals
Green runs with a major change — promotion sign-off is gated on you.
Pending approvals
| Change | Requested by | Scope | Risk | Created | Action |
|---|---|---|---|---|---|
| PORT-2830 promote v20 → v21 · API contract changed · 23/23 green |
SAGE Generate | Portal | Medium | 2 hours ago | |
| API-1142 promote v7 → v8 · new required field · 12/12 green |
SAGE AI Driver | API | High | 1 day ago |
Loading Agentic Mission Control
Reading agents, versioned skills, deterministic tools, missions, approvals, evidence, evaluations, and security posture.
Loading SAGE architecture and quality intelligence…
LLM & costs
Azure OpenAI powers the agents — control per-agent and watch spend.
All LLM
Azure OpenAI connection
connectedEndpoint—
Deployment—
Fallback when offdeterministic rules only
Tokens MTD
—
Cost MTD
—
Projected month-end
—
Budget
$2,000
Per-agent usage — MTD
projection assumes current daily rate| Agent | LLM | Tokens | Cost MTD | Projected | Share |
|---|