Scope & SLA
What Z2m Red Team covers today—and what sits outside product scope.
Safe to say
- · Dual-account IDOR/BOLA when two test users (User A and User B) are provided.
- · Confirmation gate: no customer vulnerability without replayable HTTP or OOB evidence.
- · CVSS v3.1 on confirmed findings, with redacted replay PoCs.
- · 12 first-party authenticated checks (BOLA when two test users are provided) + optional Nuclei/Prowler when those tools are installed. Nuclei/Prowler IDs are labeled separately and are not counted as Z2M playbooks.
- · Owner's-risk Metasploit auxiliary/check is not Hard Proof unless replayable HTTP or OOB evidence exists.
In scope
- · 12 first-party authenticated playbooks (BOLA needs a peer User B; race/CSRF only on optional endpoints). Nuclei/Prowler IDs are labeled when those tools run — not counted as Z2M playbooks.
- · Web, API, and AI pentest with finding triage
- · Confirmation gate: customer vulns require replayable HTTP/OOB evidence plus a PoC; Potential is not logged as a vuln unless report_mode includes investigate
- · CVSS v3.1 on confirmed findings; Hard Proof ZIP with redacted request/response. Metasploit check output is an
msf.txtappendix, not CONFIRMED. - · Agent-assisted remediation, tickets, Hard Proof exports
- · Client remediator portal, engagement reports, continuous programs
- · Org SSO, audit export, billing, and trust / RoE controls
Engagement restrictions
- · Destructive actions, denial of service, and mass targeting are disabled (fail-closed). Scan config cannot turn them on.
- · Race tests: at most 8 parallel requests, and only if you set
race_endpoint. - · Metasploit MCP stays off on shared SaaS workers. Auxiliary/check or write/session can run on a private customer agent after Trust owner's-risk consent (write needs a second phrase and a 4-hour window). Issues badge those as Owner's risk, not Confirmed, unless HTTP/OOB replay exists.
We do not claim
- · 200+ offensive playbooks
- · Zero false positives on novel zero-days
- · 150 native tools
- · Hermes reasoning engine
- · Playwright via MCP / MCP browser as a product feature
- · 70–85% token reduction as a production SLA
- · Metasploit on every authorized scan
- · Metasploit MCP on shared SaaS workers
- · Metasploit write/session on shared SaaS workers
- · Sessions or payloads as Hard Proof without HTTP/OOB
Out of scope (N/A)
- · Physical red team / social engineering engagements
- · Full cloud CSPM as a managed service (worker hooks only)
- · CPA-issued SOC 2 Type II attestation (we provide readiness evidence only)
Service levels
- · Platform uptime target: 99.9% monthly (excludes planned maintenance)
- · Support response: critical 4h · high 8h · standard 24h (business days, Enterprise+)
- · Finding SLA inside continuous programs is org-configurable (defaults: critical 24h / high 72h)
Full operational runbook: see docs/ops and the live status page.