AWARSE LABS
⚡ Zero-Trust MCP Tooling for Engineering Teams

Local-first AI agents for FinOps & QA Automation

Ephemeral, high-precision developer tools built on FastMCP. Eliminate wasted SaaS spend and flaky end-to-end tests without sending your source code or credentials off-premise.

FinOps Governance Agent

SeatPrune for GitHub & Copilot

Identify and reclaim orphaned seats across GitHub Enterprise and GitHub Copilot in seconds. Automated safety locks, dry-run simulations, and protected account governance prevent accidental de-provisioning.

  • Default --dry-run protection
  • Protected service accounts via SEATPRUNE_ALLOWLIST
  • Zero source code inspection | Strict read-only tokens
seatprune-audit.sh

# Run instant ephemeral dry-run audit

$ uvx --from seatprune seatprune-mcp

Scanning org: acme-corp

Total Seats: 182

Inactive (>60d): 24 seats identified

Potential Annual Savings: $5,040 / year

[DRY-RUN] No changes committed.

Autonomous Test Resilience

Healwright for Playwright

Autonomous Playwright test healing agent over Model Context Protocol. Inspects failing DOM snapshots, correlates UI mutations, and regenerates stable, resilient locators directly within your IDE or local Claude environment.

  • Deterministic Playwright trace & locator healing
  • Native FastMCP integration for Claude Desktop & Cursor
  • Hermetic runtime via PEP 517 wheels
healwright-agent.log

# Detecting locator failure

FAIL: locator('button.submit-v1') not found

Analyzing DOM diff snapshot...

✓ Suggestion: getByRole('button', { name: 'Confirm Order' })

Confidence: 98.4%

Enterprise Architectural Guarantees

Designed from the ground up for strict corporate security perimeters.

🔒

Zero Cloud Custody

No intermediary proxies, tracking pixels, or external databases. Your code and secrets remain inside your local perimeter.

Ephemeral Execution

Run on demand via uvx with zero system-wide pollution and zero daemon footprints left behind.

🛡️

GDPR & CCPA Compliant

Because no personal data leaves your client boundary, Awarse Labs never acts as a Data Processor under Article 28.