The QEAgency.
Built on 25 years of QE experience
Four independent, specialized agents — each owns a distinct slice of quality engineering: discovery, code, healing, and data.
Built For QE Teams — An Agency of Autonomous Agents.
An agency of autonomous agents
01How It Works
Continuous verification. End-to-end.
Three steps from a staging URL to a self-healing test suite running in your CI — change reasoning, generation, healing, on a continuous loop.
Understand the change
myNova reads requirements, code diffs, and application history. It determines what must be tested on every change — and dispatches generation tasks to the rest of the system.
Generate verification
myGenie authors idiomatic Playwright TypeScript with role and label-based locators, then commits it. myNexa aligns test data and database state with the application — in parallel.
Run, heal, repeat
Tests run on every commit. When the UI changes, myAstra recomputes selectors from the page state graph in real time — keeping suites green with zero manual maintenance.
02Platform Governance
Enterprise-grade access control.From org to agent — every layer governed.
MyDiya.AI is built for multi-organization enterprises. Onboard teams, structure projects, assign roles, and control which agents each user can access — all through a unified RBAC framework.
Organization Onboarding
Each enterprise or business unit exists as an isolated tenant — with dedicated workspaces, separate data schemas, and independent billing. Onboard organizations in minutes with zero cross-tenant data exposure.
- Schema-per-tenant isolation
- Independent workspace config
- Custom org branding
- SSO per organization
Project Management
Create and manage multiple QE projects within an organization. Each project maps to an application, a test scope, or a delivery team — with independent agent configurations and CI/CD pipelines.
User Roles & Auth
Fine-grained role-based access control ties every user to a permission profile. Roles are enforced at API, UI, and agent-execution layers — ensuring the right people have the right access, always.
Agent Assignment
Agents are assigned to users based on their role and project membership. A QE Lead has full access to all four agents; a Viewer can only read reports. Capability tokens enforce boundaries at execution time.
Role Hierarchy · Agent Access Matrix
| Role | myNova | myGenie | myNexa | myAstra | Projects | Reports | Settings |
|---|---|---|---|---|---|---|---|
| Org Admin | |||||||
| QE Lead | |||||||
| QE Engineer | |||||||
| Viewer |
Onboarding Flow · Organization to Agent in 4 Steps
Create Organization
Register your org, configure SSO or email auth, and set workspace-level policies.
Create Projects
Map projects to your applications. Each project gets its own agent configuration and CI pipeline.
Invite Users & Assign Roles
Invite team members and assign roles — Admin, QE Lead, Engineer, or Viewer — per project.
Agents Activate by Role
Agents are automatically enabled per role. Capability tokens are issued at login — no manual wiring.
03Agent Library
Four agents. Four specialties.
Each agent has a defined cognitive scope, its own model, and its own tools — and runs independently. Bring in one, or all four.
Inputs & Outputs
Each agent takes its own inputs and produces structured, ready-to-use outputs — run any one on its own.
Input
Output
- 1Optimized Steps
- 2Smart Locators
- 3Test Data
- 4API Discovery
- 5UI Bugs
- 6Screenshots
- 7Accessibility
- 8Performance Reports
04Use Cases
Four workflows these agents handle today.
What MyDiya does, exactly. No vague promises — these are the workflows the platform handles today.
Smoke suite from a URL
Point Browser Navigator at staging. It crawls the app, fingerprints page states, identifies critical user flows, and the Test Authoring agent emits a Playwright suite you can review and merge in minutes.
Record once, run forever
Hand-record a complex flow (multi-step approvals, frame-switching, downloads). MyDiya converts it to Playwright code with stable role/label-based selectors, auto-cleaning misclicks and retries.
Surviving UI churn
When a data-testid renames or a CSS class drifts, myAstra recomputes the locator from the page state graph — in real time, with no human in the loop. Maintenance debt stops compounding.
05Security
Trusted with your test estate.
Schema-per-tenant isolation, capability-scoped tokens, and audit trails — so QA leads can hand the platform to engineers without losing sleep.
Per-Agent Capability Tokens
Every agent carries a signed, attenuable token. Lock down what it can do — by workspace, project, tool, or time. Engineers can’t escalate scope.
Test-Run Safeguards
Hardware-Bound Identity
CLI keys stay in TPM / Keychain — never leave the engineer’s laptop.
Workspace Isolation
Schema-per-tenant DB isolation. Each QA team’s tests, secrets, and runs are walled off — with a full audit trail across them.
Trust Chain
Cryptographic Chain
HMAC-chained caveats
sig₀ = HMAC(rootKey, identifier)
sig₁ = HMAC(sig₀, caveat₁)
sig₂ = HMAC(sig₁, caveat₂)
... anyone can ADD a caveat, nobody can REMOVE one06Command Line
Tests from the terminal.
For QE engineers who live in the shell. Generate, run, and heal Playwright suites from the command line — and from your CI runners.
Connected · Offline
Switches automatically — full platform online, degraded on flaky CI runners, fully offline behind a firewall.
Sandboxed by Default
Every test run is sandboxed. Repo-root isolation, file-write allowlist, no shell escape.
CI-Native Output
JUnit XML, TestRail, JSON report. Plug into Jenkins / GitHub Actions / GitLab CI in one line.
Ready when you are
Put your QE on autopilot.
Point MyDiya at your staging URL and get a self-healing Playwright suite running in your CI — in minutes, not sprints.
07Help & Support
Need a hand? We're here.
Reach out and a real human from the MyDiya team will get back to you — usually within one business day.