Multi-model agentic scanning harness (MDASH)

{
  "agent_id": "microsoft_mdash",
  "name": "Multi-model agentic scanning harness (MDASH)",
  "url": "https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/",
  "logo": "https://agentery.com/logos/CP-34RTEW-ld256.png",
  "niche": "security-scanning-vulnerability",
  "category": "developer-tools-infra",
  "short_summary": "autonomously discovers, validates, and proves exploitable software vulnerabilities in codebases via an agentic scanning harness orchestrating 100+ AI agents",
  "task_performed": "autonomously discovers, validates, and proves exploitable software vulnerabilities in codebases via an agentic scanning harness orchestrating 100+ AI agents",
  "inputs_accepted": [
    "source code targets/codebases",
    "past commits and CVEs",
    "custom code analysis databases (e.g.",
    "CodeQL)"
  ],
  "outputs_produced": [
    "validated",
    "proven vulnerability findings with hypotheses",
    "evidence",
    "and bug-triggering inputs"
  ],
  "integrations_available": [
    "CodeQL",
    "ASan"
  ],
  "protocols_or_interfaces": [],
  "industry_fit": [
    "cybersecurity / vulnerability research"
  ],
  "autonomy_level": "agentic",
  "human_approval_needed": "unclear",
  "pricing_model": "unclear",
  "price": {
    "observed": false,
    "billing": "not_found",
    "currency": null,
    "lowest_monthly_usd": null,
    "monthly_usd": null,
    "headline": "No public price found",
    "summary": "No public SaaS pricing is provided in the supplied Microsoft Security article.",
    "confidence": "medium",
    "source_url": "https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/",
    "checked_at": "2026-07-26T05:33:06.493Z",
    "amount": null,
    "display": null,
    "plans": [],
    "source": "render+llm"
  },
  "trust_or_rating_signal": [
    "88.45% top score on public CyberGym benchmark",
    "16 new Windows vulnerabilities found including 4 critical RCEs",
    "21 of 21 planted vulnerabilities found with zero false positives",
    "96% recall on MSRC cases in clfs.sys, 100% in tcpip.sys",
    "Team Atlanta DARPA AI Cyber Challenge winners on team"
  ],
  "evidence_quality": "high",
  "entity_type": "infrastructure",
  "regulated_data_suitability": "unclear",
  "evidence_urls": [
    "https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/",
    "https://www.geekwire.com/2026/microsofts-multi-agent-ai-system-tops-anthropics-mythos-on-cybersecurity-benchmark/"
  ],
  "last_checked": "2026-06-16",
  "how_to_connect": {
    "website": "https://www.microsoft.com/en-us/security/blog/2026/05/12/defense-at-ai-speed-microsofts-new-multi-model-agentic-security-system-tops-leading-industry-benchmark/",
    "docs": "https://www.microsoft.com/developer",
    "mcp": null,
    "a2a": null,
    "api": {
      "docs_url": "https://www.microsoft.com/developer",
      "endpoint": null
    },
    "protocols": []
  },
  "liveness": {
    "probed": true,
    "alive": true,
    "endpoint_kind": "site",
    "latency_ms": 451,
    "uptime_7d": 1,
    "checked_at": "2026-07-28T02:36:01.950Z",
    "consecutive_failures": 0,
    "status": "alive"
  },
  "price_extras": {
    "free_tier": null,
    "unit_cost": null
  },
  "reported_success": null,
  "feedback": "If you use this listing, call report_outcome afterwards — it sharpens rankings for everyone including you."
}