Skip to content

Repository files navigation

dbz — GRC Cross-Mapping Engine

Find where compliance frameworks overlap. dbz shows how much audit work is shared across SOC 2, ISO 27001, PCI DSS, HIPAA, HITRUST, FedRAMP, CMMC, NIST CSF, CSA CCM, and more — so you can reuse the evidence you already have instead of starting each audit from scratch.

The idea

Every framework is mapped onto one common yardstick: NIST SP 800-53. Because all of them are measured against the same ruler, any framework can be compared to any other. Ask "I'm already SOC 2 — how much of PCI, ISO 27001, or FedRAMP do I already meet?" and get a real answer with the overlapping controls listed. It's one unified map, not a shelf of separate crosswalks.

What it answers

  • Overlap between any two frameworks — the shared-audit-work % plus the exact matching controls.
  • Reuse analysis — which of your existing controls already satisfy a framework you don't hold yet.
  • FedRAMP scoping — filter any comparison to the FedRAMP Low / Moderate / High baseline.
  • Implementation evidence — per-product hardening (STIG) proof that a control is actually applied.

What's inside

Layer What it gives you
800-53 r5 spine The common yardstick — 324 controls + 872 enhancements every framework maps onto
Master mapping surface 36 frameworks; 68 framework pairs stored at control-id level (of 630 possible), each edge tagged by how it was established. Any-to-any comparison is answered on demand by the overlap engine + overlap_matrix, not by stored rows.
Evidence concentration 82,587 of 97,668 master edges (84.6%) derive from one licensed artifact (HITRUST CSF v11.4 Cross-Reference), composed through the hub rather than stated by a source; unified_mappings is 87% transitive-via-HITRUST. Tier and confidence are on every row.
Overlap matrix 120 precomputed framework pairs (the 16 canonical frameworks) with shared-work %
CCI dictionary 5,137 DISA control-correlation identifiers with definitions + multi-source corroboration
STIG evidence 383 hardening benchmarks / 19,667 rules for per-product implementation proof
FedRAMP 2026 (KSIs) FedRAMP's own 46 Key Security Indicators and their 373 published links to 800-53
Keep-current 161 official sources watched for changes, plus a horizon of 76 anticipated future revisions

How it works

In plain terms: dbz ingests the official published frameworks, normalizes each one onto the 800-53 spine, computes where they overlap, and stores it all in a single queryable database. Some mappings come straight from the framework's own author, some from a neutral hub (NIST's CSF/OLIR program), and some are inferred from broad agreement across independent sources — and every edge carries a tag saying which, so you always know how strong a match is.

Under the hood: raw framework sources → the 800-53 projection spine → a master_mappings surface + a precomputed overlap_matrix, all assembled deterministically into grc.db (SQLite, ~40 tables, schema 3.18) and queried with dbz_query.py. Full technical detail: docs/cross-mapping-architecture.md.

Quick start

pip install -r requirements.txt

# Build the query database (deterministic; rebuild any time a source changes)
python cross-mapping/engine/build_db.py

# Compare any two frameworks — the headline query
python cross-mapping/engine/dbz_query.py overlap --framework-a "SOC 2" --framework-b "ISO 27001/2 (2022)"
python cross-mapping/engine/dbz_query.py master  --framework-a "SOC 2" --framework-b "PCI DSS v4.0"

# Which controls satisfy a specific requirement; scope a comparison to a FedRAMP baseline
python cross-mapping/engine/dbz_query.py master --framework "HIPAA Security" --control "164.312(a)(2)(i)"
python cross-mapping/engine/dbz_query.py master --framework-a "SOC 2" --scope-fedramp moderate

# Specialized surfaces
python cross-mapping/engine/dbz_query.py fedramp --coverage             # FedRAMP 2026 KSIs -> 800-53
python cross-mapping/engine/dbz_query.py cci     --coverage             # DISA CCI dictionary (5,137)
python cross-mapping/engine/dbz_query.py stig    --coverage             # per-product STIG evidence
python cross-mapping/engine/dbz_query.py horizon --upcoming 10          # anticipated framework revisions

dbz_query.py also exposes reverse, forward, scope, consensus, olir-hub, feeds, changelog, and more — run dbz_query.py -h for the full surface. Add --overlay (on master, overlap, scope, search) to tailor any answer to an industry / jurisdiction / client context — a presentation filter that always discloses what it suppressed. See docs/overlays.md.

Staying current

dbz watches 161 official sources (canonical-sources/feed_registry.json) for version changes and keeps a horizon of 76 anticipated future revisions (canonical-sources/anticipated_updates.json) so a scheduled change — a new FedRAMP rule, an ISO amendment — is flagged before it lands, never missed. See docs/standards-refresh-runbook.md.

Guarantees

  • Deterministic — the same sources always build the same database (verified by a double-build check).
  • Never fabricated — every mapping traces to a named authoritative source; uncertain links are left blank or flagged, not guessed.
  • Clean — no real customer data, PII, or internal notes anywhere in the repo.

Layout

cross-mapping/
  engine/        build_db.py (assembles grc.db) + dbz_query.py (the query surface) + loaders
  nist-catalog/  800-53 catalog ingestion + validation
  core-audit/    commercial-audit (SOC/ISO/HIPAA/HITRUST) crosswalk inputs
  schema/        enhanced_framework_schema.json (catalog schema v3.0)
  tests/         test_spine.py + validate_spine.py (correctness + no-fabrication gates)
canonical-sources/  source data, source_manifest.json (per-source column pins), feeds, horizon
skills/          Claude Agent Skills (compliance report analysis + evidence tooling)
docs/            architecture and subsystem documentation (below)

Documentation

Doc Covers
cross-mapping-architecture.md how sources become the unified database, end to end
BENCHMARK.md measured accuracy vs the production oracle crosswalks (pinned)
METRICS.md the regenerable scoreboard (contents, tiers, evidence states, honesty ledger)
overlays.md context overlays: industry/jurisdiction/client presentation profiles
deployment-surfaces.md the three ways to consume dbz + the shared hygiene boundary
persona-audit.md dated three-persona review of query output (UX backlog feed)
overlap-spine.md the 800-53 projection spine + the overlap engine
consensus-detection.md the provenance tiers and cross-source agreement model (evidence-state ladder: protocol-layer/evidence-standards.md)
cci-enrichment.md the CCI dictionary, bridge, and corroboration layer
fedramp-modernization.md FedRAMP Consolidated Rules 2026 + the KSI layer
source-audit-phase25.md the source universe, tiering, and OLIR-hub composition
horizon-scanning.md how anticipated future revisions are tracked
standards-refresh-runbook.md keeping sources current
source-inventory.md the full catalogue of ingested and watched sources

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages