Autonomous QA for the real world. Argus reads your requirements, scores them, generates Playwright tests, runs them on every deploy, and shows what broke — in Mission Control, a live ops console.
Requirements traced to tests · Playwright under the hood · Self-healing autofix · No LLM key for smoke · CLI · dashboard · GitHub Action · MCP
No LLM key required for smoke tests, rule-based parsing, and most dashboard actions. Add a provider when you want richer generation and analysis.
| Test intelligence | Flake taxonomy with file-backed quarantine (TTL + owner), change-based test selection from git diff and requirement links, and a failure studio per case; argus intel CLI |
| Typed model dependencies + impact canvas | Explicit edges such as Order → Payment, rendered as a typed-dependency list and SVG canvas in Mission Control |
| Requirements connectors | Jira OAuth, Gmail/IMAP fallback for email, and a diarize source for speaker-tagged transcripts |
| CRM golden-path packs | HubSpot, Pipedrive, Zoho and Salesforce, with Zoho and Salesforce UI-first (guide) |
| Ask Zyra local FastEmbed | On-box embeddings when no OpenAI-compatible embeddings endpoint is available |
| Desktop remote Mission Control | The Tauri shell opens a lab or team argus serve by URL |
Full history: CHANGELOG.md.
| When this happens… | Argus gives you… |
|---|---|
| Specs and tests drift apart | Requirements that are versioned, scored, and traced to every generated test |
| Smoke runs are a tribal ritual | One command, or one dashboard click, on every deploy |
| Flaky selectors waste the afternoon | Autofix suggests a repair and re-runs |
| API drift is tribal knowledge | OpenAPI contract test, breaking-change diff, and HAR consumer verify |
| Security checks live in spreadsheets | Authorized misconfig, CVE, SCA, DAST, LLM red-team and chaos jobs with an audit trail |
| QA tooling is five products | One Mission Control: Console, Testing, Security and Operations |
- Requirements — versioned, scored, and traced to every generated test, plus impact by shared data models and flows
- One pipeline — the same LangGraph path from the CLI or one dashboard click
- Self-healing autofix — suggests and applies repairs, then re-runs
- Contracts — OpenAPI contract test, breaking-change diff, and HAR consumer verify
- Authorized security — misconfig, CVE, SCA, DAST, LLM red-team, and chaos jobs with an audit trail and sandboxed PoC
- Mission Control — Console, Testing, Security, and Operations in one rail
Ask Zyra (optional knowledge extra): docs/tutorials/14-ask-zyra-knowledge.md. What DAST covers, and what it deliberately defers: docs/security-network-attack-gaps.md.
| Argus | Hand-written Playwright suite in CI | |
|---|---|---|
| Test runner | Playwright | Playwright |
| Where tests come from | Generated as .spec.ts files from versioned, scored requirements |
Written and maintained by hand |
| Spec traceability | Every generated test linked to a requirement version | Kept by convention, if at all |
| Broken selectors | Autofix suggests and applies a repair, then re-runs | Fixed by hand |
| Flaky tests | Flake taxonomy, quarantine with TTL and owner, change-based selection | Retries and manual triage |
| API contracts | OpenAPI contract test, breaking-change diff, HAR consumer verify | Separate tooling |
| Security checks | Authorized, engagement-gated jobs with an audit trail | Separate tooling |
| Results | Mission Control console with live job terminal | HTML report per run |
| Choose plain Playwright when | You have a small, stable suite your engineers already own and no need for requirement tracing |
flowchart LR
subgraph Sources["Specs"]
GH["GitHub"]
PDF["PDF"]
Email["Email"]
Jira["Jira"]
Transcript["Transcript"]
end
Sources --> Pipeline["LangGraph pipeline"]
Pipeline --> Playwright["Playwright"]
Playwright --> Console["Mission Control"]
Pipeline -->|"fail"| Autofix["autofix"]
Autofix --> Playwright
argus test run --source github --spec docs/specs/feature.md
argus test run --source document --spec requirements/checkout.pdf
argus flow run https://zyvor.dev --steps docs/assets/zyvor-dev-demo.steps --videoCommand reference: docs/test-authoring.md.
Requires Python 3.11+, Node 20+, and Docker (only for the container path). make install handles the rest, including Playwright's Chromium download.
git clone https://github.com/zyvorai/zyvorai-argus.git && cd argus
cp .env.example .env # set ZYVOR_BASE_URL
make install # Python venv + Playwright Chromium
argus test exec --grep @smoke # first green run — no API key
argus serve --port 8080 # → http://localhost:8080/dashboardRemote deploy in one command:
./scripts/deploy-remote.sh YOUR_HOST YOUR_USER --service --key
# Mission Control on port 30080 — credentials printed in deploy summaryContainer:
docker pull ghcr.io/zyvorai/zyvor-argus:v0.9.2
docker run --rm -p 8080:8080 --env-file .env ghcr.io/zyvorai/zyvor-argus:v0.9.2 serve --port 8080 --host 0.0.0.0| Track | Where |
|---|---|
| Non-production use (free under the Zyvor Production License) | This repo |
| Production / commercial license | https://zyvor.dev · Buy Professional (1 year) |
| Docs | Tutorials · zyvor.dev/docs |
More: user manual · feature guide · configuration · remote deploy · enterprise overlay.
argus serve exposes Mission Control at /dashboard.
- Grouped side rail — Console (Overview, Ask Zyra) · Testing (Pipeline, Visual, Quality, Journeys, API, Probes, Requirements) · Security · Operations (Runs & schedules)
- Header — knowledge lamp, dark/light theme, Search (also ⌘K)
- Overview — hero status, pass rate, next smoke, and a live macOS Terminal job panel (Copy / Save / Stop)
- Hash routes —
#pipeline,#ask,#requirements,#operations
Full action list: dashboard tutorial.
- Any CI/CD pipeline — the
action.ymlGitHub Action runsargusPlaywright-based QA checks against any target URL (external CI/CD tutorial). - Chat-ops — an MCP server exposes a subset of Mission Control's
/api/v2job API to MCP-capable chat agents. - Kubernetes — manifests under
kubernetes/(CI/CD and Kubernetes tutorial).
What's free under the Zyvor Production License vs. what needs a commercial license (full guide):
| Use case | Allowed without a paid license? |
|---|---|
| Development, testing, evaluation, research, education | Yes |
| Non-production laboratory and proof-of-concept use | Yes |
| Production environments and customer workloads | No — needs a commercial license |
| SaaS, managed services, OEM, appliances | No — needs a commercial license |
| Redistribution or resale | No — needs written permission and a commercial license |
Current release: v0.9.2 (releases). What is shipped and what is deliberately deferred is tracked in ROADMAP.md and the CHANGELOG.
| Product | Role next to Argus |
|---|---|
| Argus | Autonomous QA: requirements to Playwright tests, autofix, contracts, authorized security jobs |
| Chimera | Infrastructure protocol simulator; pairs with Argus when the thing under test talks to vSphere, Nutanix, Hyper-V, AWS or Azure |
| Aurora | AI go-to-market platform; sits next to Argus in the Zyvor AI products |
| Aether | Runtime portability plane for the apps Argus tests |
Argus is source-available under the Zyvor Production License v1.0 (SPDX LicenseRef-Zyvor-Production-1.0).
- Free for development, testing, evaluation, research, education, and non-production labs.
- Production use (production, customer workloads, SaaS, managed services, OEM, redistribution, and other revenue-generating use) requires an annual enterprise subscription. Plans, support levels and terms: docs/SUBSCRIPTION-MODEL.md · licensing guide · Pricing · sales@zyvor.dev.
Commercial terms are issued separately: https://zyvor.dev.
Contributions: CLA.md + DCO.md (git commit -s), governed by our Code of Conduct; see CONTRIBUTING.md. Report vulnerabilities privately per SECURITY.md.
