The Repository-Native QA Layer Across the Entire SDLC & STLC. Eliminating disconnected test silos with five unified, modular products designed for modern development teams and AI coding agents.
From autonomous test creation and Git-native management to runtime self-healing and cryptographic governance, each product fulfills an exact, mission-critical role in your repository.
The engine and tools coding agents use to drive browsers, verify web applications, generate automation code, and run developer PR checks.
tamash lint duplicate checktamash-mcp protocol server for Claude Code & Cursornpx @vibetestq/tamash-agent run --scenario features/auth.md
The repository index and Web UI for living requirements, scenarios, review status, manual test runs, tracker synchronization, and Git branching parity.
.tamash/, features/, data/traceability.json requirement-to-code mapnpx @vibetestq/tamash-qe init && npx tamash-qe ui
The autonomous runtime library in every test run that intercepts broken locators on the fly, logs heals, and synthesizes clean AST code Pull Requests.
tamash-heal.json with @TC-... IDsapply-heals creates clean Git PRs with strictly zero XPathimport { test } from '@vibetestq/tamash-playwright';
Native reporters recording results, healed steps, and failure analysis per test case; report views integrate directly into the Git repository and Web UI.
traceability.jsonreporter: [['@vibetestq/tamash-playwright-dashboard']]
Enterprise security and governance engine providing Ed25519 cryptographic license verification, seat management, SSO, and audit logging.
TAMASH_LICENSE_KEY="ed25519.sig.ey..." npx tamash-qe check
When an application UI shift causes a locator to time out, tamash-playwright intercepts the exception in milliseconds, executing a prioritized, multi-model healing cascade:
Matches the ARIA tree, roles, accessible names, and aria-describedby relationships. Instant sub-10ms resolution.
Analyzes visual container boundaries, spatial geometry, and relative anchor coordinates to locate the target element.
Computes Levenshtein distance across button text variants, casing differences, and placeholder modifications.
Applies localized semantic judgement when confidence is ≥ 0.80. Safely falls back if ambiguity remains.
Zero False Passes Allowed
TAMASH heals user actions (clicks, inputs, selections) to navigate through cosmetic UI changes, but NEVER alters test assertions (expect(...)). If the business rule fails, the test reports a genuine defect.
Unlike temporary in-memory hacks, TAMASH produces permanent code maintenance:
How the 5 TAMASH products operate as a unified system across every phase from user story inception to production maintenance.
tamash-qe imports user stories from Jira, Azure DevOps, or GitHub Issues. tamash-agent analyzes acceptance criteria and drafts plain-English scenarios verified by tamash lint.
Developers or AI coding agents crawl the feature branch. tamash-agent generates verified Playwright specs and locator factories stored directly in the Git repository alongside feature code.
Testers execute manual checklist runs inside the tamash-qe Web UI while tamash-agent automates repetitive login and setup preconditions.
tamash-agent uses git diffs to run only affected tests. tamash-playwright intercepts broken locators on the fly to prevent pipeline blocking.
Heals logged in tamash-heal.json trigger apply-heals, creating a clean Pull Request with updated locators without touching test assertions.
tamash-qe compiles release readiness matrices from traceability.json, while tamash-admin-portal enforces air-gapped cryptographic compliance.
Eliminating fragmented databases, manual locator maintenance, and costly dedicated servers with plain files in Git.
| Capability | Traditional Siloed QA Stack | TAMASH Repository-Native Layer |
|---|---|---|
| Test Case Repository | External database (TestRail, Zephyr) disconnected from Git feature branches | Git Branching Parity: Scenarios live as Markdown/Gherkin in repository |
| Locator Maintenance | Manual fixes consume 30-40% of every sprint; CI pipelines constantly blocked | 4-Tier Autonomous Healing: Runtime interception + AST Git PR generation |
| Assertion Integrity | Heuristic tools often generate false passes or mask product defects | Non-Invasive Assertion Guarantee: Heals actions only; 0 false passes |
| Reporting Infrastructure | Dedicated servers (ReportPortal, Allure Docker, Elasticsearch) costing $30k+/yr | Zero-Server Reporting: Cross-run reports built as lightweight artifacts |
| AI Coding Agent Ready | Incompatible; AI agents cannot authenticate to proprietary web UIs | Native MCP Server: Claude Code, Cursor, and Windsurf drive browsers directly |
| Security & Licensing | Telemetry phone-home required; data exposed to external cloud multi-tenants | 100% Air-Gapped: Ed25519 asymmetric cryptographic license validation |
Integrate TAMASH Self-Healing into your existing Playwright repository with a single npm package, or deploy the complete Git-Native QE layer across your engineering organization.