NSAuditor AI is open core, runs offline, and ships with a typed plugin SDK. These docs cover the install, the architecture, the plugin model, the compliance evidence, and the air-gapped operation guide that procurement asks about.
CURRENT · EE 0.33.1 · 2026-08-07 EE 0.33.1 · CE 0.2.38 · agent-skill 0.2.36
The wiring release — eleven compliance options that nothing shipped could populate now receive values. They were reachable only in theory: the option existed, the engine read it, and no entry point ever wrote to it. The new entry points live in Community Edition, so EE 0.33.0 requires CE ≥ 0.2.37 — on an older CE the options stay unfed. Alongside it, the Zero-Data-Exfiltration claim was rebuilt as a positive register of 17 enumerated outbound paths rather than a blanket denial, so a reviewer can read what the process does contact instead of taking “nothing” on faith; and signature records now carry algorithm and backend frozen at the moment of signing, not re-derived later from whatever the current configuration happens to be. RFC 3161 trusted timestamping is now opt-in via NSAUDITOR_TSA_URL and was verified against a real timestamp authority on 2026-08-07 — it is opt-in, not a default, and it is an outbound call. Ed25519 suppression signing is still not reachable — no shipped entry point signs a suppression record. Matrix-neutral — all seven coverage matrices UNCHANGED and the plugin count UNCHANGED at 28 Enterprise plugins (27 of them cloud auditors), 55 overall. Paired CE 0.2.37 + agent-skill 0.2.35.
Previously — EE 0.32.11 (2026-08-05). The dependency-advisory gate was auditing the wrong tree. It ran its audit inside the maintainer’s development checkout and called the result the production closure. It now packs the tarball, installs it the way a customer does, and audits that. With both readings taken side by side: the development tree carried 25 advisories, 8 of them HIGH; the closure a customer installs carried 6, none HIGH. The two lists share 5 of 26 packages and zero high-severity ones — every one of the eight highs was development-only. A matching total is exactly what makes a wrong list read as right, so the gate now reconciles the two as sets and prints the package-for-package divergence on every run.
And it refuses to report clean until an advisory database has answered. With the audit taken offline the gate produced a well-formed zero — dependency counts non-zero, the summary reconciling against its own list — and printed a pass over a closure that really carried six. It now installs a pinned canary carrying a permanently published advisory and requires that advisory back; silence is an untrustworthy run, not a pass. It also prints which registry answered, because a canary proves liveness, not freshness.
SOC 2, enumerated in full — 10 covered / 4 partial / 37 out of scope = 51. Four AICPA criteria (CC7.4, CC7.5, A1.1, A1.3) appeared nowhere in the mapping data: silently non-enumerated, which is not the same thing as out-of-scope-with-a-reason. Each is now enumerated with an architectural reason and the operator-side evidence that closes it. No control changed status and no routing changed — this is enumeration completeness, not a coverage bump — and the other six matrices are unchanged. Separately, the Type II tables now state which mechanisms are reachable from a shipped entry point instead of presenting all of them as supported. 28 Enterprise plugins — 27 of them cloud auditors — 55 in total, the same seven frameworks — unchanged. Live-verified on the published bytes: 28 of 28 Enterprise plugins active, and a three-cloud scan wrote a 76-file evidence pack per cloud.
Previously — EE 0.32.9 (2026-07-29). An internal-provenance strip: a compliance report is a document a customer hands their auditor, and ours carried internal engineering identifiers — roadmap ids, internal release stamps, the name of an internal audit review — in finding titles, on the attestation cover page and in the chain-of-custody record; measured on a rebuilt three-cloud evidence pack, 686 unexplained internal-marker occurrences → 0 across 105 files, with a positive control in the same run (3,572 benign matches still detected). And a false-clean closure: when a cloud plugin could not start — optional SDK absent, credentials unusable — it refused to report, and that refusal evaporated one layer up, so the report came out byte-identical to one where the scanner ran and found nothing and ten controls read PASS. Every in-scope control of a cloud that could not be scanned now carries a fail-closed evidence gap, and an archived scan re-processed by a newer build WARNS instead of failing clean. Community Edition 0.2.34 fixed --out <dir> writing to the parent directory when the directory name contains a dot, made the MCP scan_cloud summary handle both spellings of the evidence-gap prefix, and stopped validate misreporting where plugins came from.
Upgrading — three things to know. (1) Coming from a build older than 0.32.9: the evidence-gap finding title changed, so suppression rules written against the old text stop matching. The direction is safe — findings resurface rather than hide — but it is silent, so re-check your suppression list against the new titles. (2) Re-scan rather than re-processing scans captured before 0.32.9; the engine warns when it is handed an older archive. (3) Your next SOC 2 report carries four additional out-of-scope rows — CC7.4, CC7.5, A1.1 and A1.3, each with its reason. Nothing you were passing has moved.
From zero to first scan in under two minutes — no signup, no credit card, no telemetry. Works on macOS, Linux, and Windows.
New customer? The linear path from your purchase email (license key + npm token) to your first audit-evidence report: install Node.js + the trio, activate your license, configure AWS/Azure/GCP credentials, run your first cloud audit, scope it with --aws-region (one region, a CSV list, or every enabled region), and drive it conversationally from Claude Desktop via MCP.
Install in 30 seconds. Run your first scan, decode the output, point it at AWS / TLS / DNS / OT targets, export reports.
github · README →How the engine, plugins, AI providers, and report writers fit together. Where data flows. What stays local. Why nothing leaves.
github · README →All 55 plugins (27 network/host + 28 Enterprise — 27 cloud auditors plus 1023 Zero Trust Assessment, which scores a network-host scan rather than calling a cloud API) listed with what they probe, what they verify, false-positive notes, and the YAML manifest of controls each one satisfies. AWS auditors include 1070 KMS, 1080 Lambda Security, 1090 Secrets Manager + SSM, 1100 CodePipeline + CodeBuild, 1110 IAM Effective Decrypt-Path, 1120 S3 Lifecycle + Replication, 1130 AWS Backup (12-dimension air-gapped vault attestation — the A1.2 ransomware-defense substrate), 1140 RDS + Aurora, 1150 SQS + SNS, 1160 VPC Endpoints, 1170 EC2 Security-Group Perimeter, 1180 ElastiCache/Redis, 1190 SES, 1200 Inspector + GuardDuty, and 1210 EC2 Instance Metadata.
github · plugins/ →How the offline NVD feed works, how version-string matching turns service banners into CVE findings, and how risk scoring plus the operator suppression workflow (accepted-risk / false-positive) prioritize them. Findings are emitted unverified — no automated probes.
github · README →Run NSAuditor AI as an MCP server. Wire it into Claude Code, Cursor, or any MCP-aware client. 5 CE tools, 9 Pro tools, and a Skill package for agents.
github · README →AICPA TSC 2017 mapping — 10 covered + 4 partial + 37 explicitly out-of-scope across the complete 51-criterion universe, with multi-cloud evidence (AWS S3 + IAM, Azure RBAC + NSG + Storage, GCP IAM + Cloud Storage, on-prem network). C1.2 Disposal with S3 Object Lock COMPLIANCE-mode. Cover-page Scope Attestation, SHA-256 chain-of-custody, a documented suppression workflow, GRC-ready evidence JSON (Vanta connector library). Type I & Type II.
read the SOC 2 guide →HIPAA §164.312 Technical Safeguards mapping — 7 covered + 3 partial + 45 explicit OOS (entire §164.308 Administrative + §164.310 Physical with named architectural-limit reasons). Per-control Required/Addressable discipline + HHS rule-text verbatim. §164.312(c)(1) ransomware-defense substrate via aws-backup-auditor Logically Air-Gapped Backup Vault cross-verification (HHS-OCR 2024 enforcement-relevant). Dual-framework one-scan workflow: --compliance soc2,hipaa. Zero BAA required — ePHI never leaves customer infrastructure.
NIST Cybersecurity Framework 2.0 (NIST CSWP 29, February 2024) mapping at the auditor-canonical Subcategory level — 13 covered + 10 partial + 83 OOS across 106 of CSF 2.0's 107 Subcategories. Govern function OOS-by-design (GV.SC-04 partial as substrate exception). Respond function OOS-entirely (IR runbook execution). Implementation Tiers 1-4 OOS as organizational-maturity claims — cover-page disclaimer (markdown + HTML parity). Triple-framework one-scan workflow: --compliance soc2,hipaa,nist-csf produces three complete evidence packs. Air-gapped operation for DFARS / CMMC / federal-contractor threat model.
PCI DSS v4.0.1 (PCI SSC, June 2024 errata; v3.2.1 retired March 31, 2024) mapping at the auditor-canonical sub-requirement level — 19 covered + 9 partial + 39 OOS across 67 of PCI DSS's ~250 sub-requirements (MVP-67). Req 12 Information Security Program OOS-by-design entirely (governance, policy, Targeted Risk Analysis Req 12.3.1 Defined-only, Customized Approach Documentation Req 12.3.2 Defined-only, TPSP Responsibility Matrix Req 12.8.5 Defined-only, IR program). Req 5 anti-malware + Req 9 physical OOS-entirely (endpoint EDR + facility-tier). Defined-vs-Customized Approach discipline per Appendix E — 15 Defined-only sub-requirements enforced at schema layer. CHD Scope operator-attested via CDE Data Flow Diagram per Req 1.2.4 + Req 12.5.1. CAO authorship per Appendix D populated on every customized-eligible entry. Card-brand AOC enforcement view (Visa CISP / Mastercard SDP / Amex DSOP / Discover DISC).
read the PCI DSS guide →ISO/IEC 27001:2022 (ISO + IEC, October 2022; 2013 edition retired October 31, 2025) mapping at the auditor-canonical per-Annex-A-code level — 17 covered + 14 partial + 62 OOS across 93 Annex A controls (the complete Annex A universe across 4 themes: A.5 Organizational 37 + A.6 People 8 + A.7 Physical 14 + A.8 Technological 34). Statement of Applicability per Clause 6.1.3.d discipline — every control carries soaApplicability field; engine produces substrate for INCLUDED controls; SoA inclusion/exclusion is operator-side. ISMS Management-System Clauses 4-10 OOS-by-design with 7 Major Nonconformity classes — absence of internal audit per Clause 9.2 or management review per Clause 9.3 = auto-fail Stage 2 (the most-frequent first-time certification failure mode). 11 NEW 2022 controls explicitly enumerated (3 COVERED + 2 PARTIAL + 6 OOS): A.5.7 Threat intelligence · A.5.23 Cloud services · A.5.30 ICT readiness for BC · A.7.4 Physical security monitoring · A.8.9 Configuration management · A.8.10 Information deletion · A.8.11 Data masking · A.8.12 DLP · A.8.16 Monitoring activities · A.8.23 Web filtering · A.8.28 Secure coding. 5-attribute taxonomy NEW in 2022 (controlType / informationSecurityProperties / cybersecurityConcepts — 5 categories, NOT 6 like NIST CSF 2.0 / operationalCapabilities / securityDomains). 2013-to-2022 transition discipline enforced at schema layer (35 unchanged + 23 renamed + 57 merged-into-24 + 11 NEW = 93). Seven-framework one-scan workflow: --compliance soc2,hipaa,nist-csf,pci-dss,iso-27001 produces five complete evidence packs. Pair with ISO-aware GRC (Drata ISO 27001 / Vanta ISO 27001 / AuditBoard / OneTrust ISMS / Secureframe ISO 27001) for SoA + internal audit + management review workflow.
CIS Controls v8 (Center for Internet Security, May 2021; v8.1 errata June 2024) mapping at the per-Safeguard level — 17 covered + 23 partial + 113 OOS across 153 Safeguards / 18 Controls (engine substrate IG1 23-of-56 / IG2-cumulative 38-of-130 / IG3-cumulative 40-of-153). CIS-Hardened-Image detection is LIVE across AWS + Azure + GCP (plugin 1210 aws-ec2-instance-auditor + Azure/GCP image-inventory producers; the renderer flips the credit from eligible → observed when a Hardened-Image is detected) + Safeguard 9.5 Implement DMARC partial. A dedicated Azure Storage Account data-protection auditor (plugin 1220) covers blob recoverability (→ CIS 11.1) + per-container public-access (→ CIS 3.3) dims, and a dedicated Azure NSG perimeter auditor (plugin 1221 — the Azure analog of AWS 1170, routing to CIS 4.4/12.2/4.2) rounds out the perimeter coverage. Implementation Group cumulative discipline — IG1=56 (the cyber-insurance baseline; ~50-70% of mid-market policies require IG1 attestation), IG2 cumulative=130, IG3 cumulative=153; smallest-IG-membership tagging (NEVER report IG2 as 74-of-74 in isolation; the IG1 base must be intact before any IG2/IG3 claim). No-certification-body attestation discipline — CIS has no formal certification body; engine output is INPUT to CSAT / CIS-CAT Pro self-attestation OR a SOC 2 auditor cross-validating CIS scope OR CIS-SecureSuite peer review, NEVER "CIS certified." Cloud Companion Guide v8 shared-responsibility-model boundary per Safeguard + CIS-Hardened-Image substrate-evidence credit (Safeguards 4.1/4.2/4.6). 5 Security Functions (NOT 6 — no Govern) + 6 Asset Types + MS-ISAC/EI-ISAC/H-ISAC sector baselines + v7.1-to-v8 cross-reference. Hepta-framework one-scan workflow: --compliance soc2,hipaa,nist-csf,pci-dss,iso-27001,cis-v8,gdpr produces seven complete evidence packs (the seventh being the GDPR Article 32 security-of-processing infrastructure substrate). Pair with CIS-aware GRC (Drata CIS / Vanta CIS / AuditBoard CIS) or CIS-CAT Pro for the self-attestation workflow.
GDPR Article 32 infrastructure substrate — security of processing. 4 covered + 5 partial + 2 OOS across 11 sub-measure units. Art. 32 substrate only, NOT GDPR compliance; Art. 83(4) lower fine tier.
read the GDPR Article 32 guide →Push your compliance-scan evidence to your GRC platform at scan time. Set COMPLIANCE_GRC_PROVIDER + COMPLIANCE_GRC_TOKEN and each scan maps its findings to Vanta test results, Drata Custom Connections records, or Secureframe evidence-collection records and pushes them — opt-in, Zero-Data-Exfiltration by default, token never serialized. Suppression-aware outcomes, deterministic idempotency, rate-limit backoff, circuit breaker, and a per-control audit artifact. Vanta activated; Drata shipped; Secureframe shipped (early-access). Early-access, single-workspace; live-tenant validation in progress.
The 6-page brochure customers and resellers ask for: what NSAuditor AI Enterprise does, how one scan produces seven framework-mapped evidence packs (SOC 2 · HIPAA · NIST CSF 2.0 · PCI DSS v4.0.1 · ISO/IEC 27001:2022 · CIS v8 · GDPR Article 32 substrate), the full Enterprise capability set (cloud scanners, GRC platform connectors, air-gapped operation, Zero Data Exfiltration, continuous monitoring), current coverage counts, edition comparison with tier pricing, and the Reseller Partner Program. Letter format, ~110 KB — built to forward to procurement, security review, and partners.
download the brochure (PDF) →A practical, vendor-neutral walkthrough of the complete audit lifecycle from the team that has built network security auditors since 2004: what a network security audit is, the role of a network auditor, the 10-step audit checklist (scope & inventory → discovery → configuration review → vulnerability validation → IAM → DNS/email → cloud accounts → logging/backups → compliance mapping → remediation & re-audit), manual vs. automated auditing, audit frequency, and best practices — including read-only audit credentials and keeping evidence on your own infrastructure. Embeds the 6-minute "Network Security Audits" explainer video.
read the audit guide →Walk-through of a hepta-framework scan against a fictional Acme Corp AWS account 111122223333. One scan, 11 AWS EE plugins, 76 findings routed across SOC 2 AICPA TSC 2017 + HIPAA Security Rule §164.312 + NIST CSF 2.0 + PCI DSS v4.0.1 (MVP-67) + ISO/IEC 27001:2022 + CIS Controls v8 + GDPR Article 32. 69 evidence artifacts per scan with SHA-256 chain-of-custody. Includes PCI DSS deep-dive showing 4 load-bearing schema enrichments + 15 Defined-only Appendix-E sub-requirements + CDE-scope operator-attestation disclaimer. All data synthetic — no real infrastructure disclosed.
view the sample scan →A linux/amd64 container image or an npm host install, with signed JWT licensing validated locally — no phone-home and no heartbeat, tcpdump-verifiable. The scan path makes no outbound calls once NSAUDITOR_OFFLINE_ONLY=1 is set; AI enrichment and the GRC push are the only other outbound paths, and both are opt-in and off by default.
Write your own plugin in TypeScript. Typed signature, scaffolder, local test harness. Same SDK we use to ship the 55 built-in plugins.
github · sdk/ →Version history, breaking changes, security advisories. Subscribe to the RSS feed of releases or watch the repo on GitHub.
github · releases →Install via npm, run your first scan, browse the plugin source, write a custom plugin. Source-of-truth lives on GitHub.
→ github.com/nsasoft/nsauditor-aiRisk-scored prioritization and an operator suppression workflow (accepted-risk / false-positive) cut false-positive triage. CVE matching uses an offline NVD feed. Reports export as PDF, JSON, SARIF, or HTML.
→ NSAuditor AI ProMSA, DPA, net-30 invoicing. Air-gapped operation for OT, federal-contractor (DFARS / CMMC), payment-processing CDE-isolation, ISO 27001 ISMS-scope-controlled, and regulated environments. Hepta-framework auditor-grade evidence — SOC 2 + HIPAA §164.312 + NIST CSF 2.0 Core + PCI DSS v4.0.1 + ISO/IEC 27001:2022 + CIS Controls v8 + GDPR Article 32 security-of-processing substrate — with SHA-256 chain-of-custody and GRC-ready evidence JSON (Vanta connector library). Zero BAA required (ZDE).
→ SOC 2 guide · HIPAA guide · NIST CSF 2.0 guide · PCI DSS guide · ISO 27001 guide · CIS v8 guide · GDPR Art. 32 guide →Try it free with the MIT-licensed Community Edition. Upgrade to Pro for CVE matching and risk-scored prioritization, or to Enterprise for cloud plugins and compliance frameworks.