Skip to content

Unified Audit Trail, Service Log, and Operator Console Architecture ​

Resolving Verdict Snapshot v3 Pointers ​

When an operator opens a governed transaction in the CaberUserConsole, the Express Backend For Frontend (BFF) joins the OpenSearch event metadata (csi_agg_events / csi_audit_unified) with the content addressed Verdict Snapshot v3 blob in GSS.

  1. Pointer Extraction: The console reads audit.edge_verdict_snapshot_state, audit.edge_verdict_snapshot_ref (for single chunk snapshots), or audit.edge_verdict_snapshot_chunks plus audit.edge_verdict_snapshot_sha256 (for multi chunk snapshots) from the OpenSearch event document.
  2. Cryptographic Verification: After streaming the snapshot from GSS via caber_storage, the BFF verifies that the payload digest matches audit.edge_verdict_snapshot_sha256.
  3. Source Coordinate Overlay: The console renders the redaction_decisions array from the v3 snapshot directly over the source payload. Because every RedactionDecision anchors its source_spans (SourceSpan { start, end }) to coordinate_space: "source", the UI can simultaneously highlight:
    • Enforced Redactions (mode: "enforced"): Byte spans that were replaced on the wire, displaying the winning policy, driver_nmers, replacement string, and identity minimized reason.
    • Simulated Observe Decisions (mode: "would_have"): Byte spans from policies or hosts running in observe mode (redactEnabled = false or redactStrategy: "disabled"), allowing compliance teams to preview exact blast radius before switching a policy to enforce.

3. Operator Console (CaberUserConsole v3.2.0) Workspaces ​

3.1 Console Route Map and Functional Proof Surfaces ​

The compiled operator guide (release-docs-generator) maps each CaberUserConsole route to a concrete functional verification surface:

Console RouteWorkspace ViewFunctional Proof & Telemetry Surface
/admin/homeExecutive Governance OverviewDisplays fleet wide governed request volume, active enforce vs observe policy posture, redaction counts, and top contributing CORI sources.
/admin/auditAudit V2 Four Zone WorkspaceFour zone inspection layout: (Zone 1) Time & Verdict Filter Bar, (Zone 2) Paginated Event Stream (csi_agg_events), (Zone 3) Action Trace Waterfall (trace.trace_id), and (Zone 4) Inline Payload & Verdict Snapshot v3 Diff Inspector.
/admin/audit?traceId=<id>Trace Detail Slide InShows freshness, provenance, and policy together on each retrieval step: a stale version flagged by its source and withheld from the agent only when the Stale metapolicy enforces, the source path behind every result linked to the Duplicate Tracker, and the determining policy with its deny or redact verdict. Adding &governance=open opens the governance decision panel on load.
/admin/overlappingDuplicate Data Tracker & Side by Side DiffVisualizes cross repository duplicate clusters, 99% capped overlap edges between documents, convex policy hulls (enforce vs observe conflict zones), and side by side chunk comparison between fresh canonical documents (version_sequence == 0) and stale replicas.
/admin/accessAuthPrincipal & Credential Authorization MatrixDisplays resolved iss_sub principals, SCIM directory groups (csi_identity_map), caber.email four step fallback hits, and masked cred_hmac keys.
/admin/dataFlowData Lineage Sankey DiagramInteractive Sankey flow mapping byte volume and chunk propagation from origin CORI storage buckets through RAG retrievers, MCP servers, and outbound API hosts.
/admin/discoveryClosed API & Shadow AI InventoryLive catalog of ApiMapMinter discoveries across the five closed resource types (http-api, mcp, object-store, bash, script) with first called and last called timestamps.
/admin/config/policiesContract v3 Policy StudioAuthoring, priority ordering, and observe_all dry run simulation for system, framework, data_use, and object_auth_metapolicy Rego bundles, contrasted against coarse over blocking (100% document deny) and regex under blocking failure modes.
/admin/contextGraphInteractive Neo4j Context Graph ExplorerForce directed neighborhood explorer for :Data, :Region, :Sentence, :LNKD, :Principal, and :API nodes, highlighting LNKD-{xor}-{count} equivalence classes and "Junk" boilerplate hubs.
CONSOLE FUNCTIONAL PROOF/admin/auditAudit Trace Detail: Freshness, Provenance and Policy on One Retrieval Step

Assertion Verified: Proves that the console shows freshness, provenance and policy together on a single retrieval step: the evidence strip counts fresh, stale and source results, a stale chip names the source file that is out of date, a provenance chip links every source of the step to the Duplicate Tracker, and the governance panel names the determining policy, its deny verdict and which metapolicies acted.

Stale result named by its sourceEvery source linked through the provenance chipDetermining policy and deny verdictFresh, Stale and Sources counts on the step
Caber Console trace detail at /admin/audit with the governance panel open on one retrieval step, whose evidence strip reads Fresh 14, Stale 1 and Sources 15 beside the determining policy and its deny verdict, with a stale chip naming its source file and a provenance chip linking all 15 sources (Light mode)
Caber Console trace detail opened from /admin/audit with the governance panel expanded, showing on one retrieval step how many results were current or stale, which source was stale, where the results came from, and which policy decided what the agent received.
CONSOLE FUNCTIONAL PROOF/admin/overlappingDuplicate Data Tracker & Side-by-Side Freshness Cluster Inspector

Assertion Verified: Proves that Caber detects 85% to 99% enterprise content duplication down to 40-byte WRL sub-sentence phrases (97% to 98.7% coverage), separates canonical fresh sources (version_sequence == 0) from stale replicas, and enforces origin trade-secret policies even when text is copied into unlabeled channels.

85%–99% Enterprise Duplication Tracked97%–98.7% WRL Sub-Sentence Coverage (40B min)99%-Capped Overlap Edges & Policy HullsFresh (seq=0) vs Stale Replica Resolution
Caber Console /admin/overlapping showing 85% to 99% enterprise duplicate clusters, 99% capped overlap edges, and side-by-side fresh vs stale source comparison (Light mode)
Caber Console Duplicate Data Tracker (/admin/overlapping) comparing a canonical fresh source against an unlabeled stale replica across shared 40-byte WRL spans.
CONSOLE FUNCTIONAL PROOF/admin/config/policiesContract v3 Policy Studio & Verdict Snapshot v3 Diff Inspector

Assertion Verified: Proves that Contract v3 Rego/Wasm policies and multisource metapolicies surgically redact only the unauthorized byte spans ([start, end) in source coordinate space) while preserving 94% of safe surrounding context in the response.

Contract v3 Wasm Rego Evaluation94% Safe Surrounding Context PreservedEnforce vs Observe (would_have) Simulation+100µs Dual-Event Wire Verification
Caber Console /admin/config/policies showing Contract v3 Rego/Wasm metapolicy rules alongside Verdict Snapshot v3 inline span redaction (Light mode)
Caber Console Policy Studio (/admin/config/policies) pairing Contract v3 metapolicy rules with side-by-side Verdict Snapshot v3 span redaction proof.
CONSOLE FUNCTIONAL PROOF/admin/contextGraphNeo4j Context Graph & Ontological Provenance Explorer

Assertion Verified: Proves that identical sentence bytes in two different documents carry distinct meaning and governance verdicts depending on their ontological neighborhood (:Region NER rollup, LNKD-{xor}-{count} equivalence class, and CORI provenance).

Content is Finite; Context is InfiniteMorgan Company Ambiguity DisambiguatedLNKD-{xor}-{count} Equivalence ClassesDecision E7 :Region NER Rollup
Caber Console /admin/contextGraph showing identical sentence bytes governed differently based on Investment Banking vs Public SEC Filing provenance and ontology (Light mode)
Caber Console Context Graph Explorer (/admin/contextGraph) proving how ontology and provenance disambiguate identical sentence content across Morgan's company divisions.

3.3 Operator Dashboard REST APIs ​

The Caber Operator Console and administrative automation interact with the coordinator through dedicated REST endpoints:

  • GET /v1/config/apimap/{host}: Retrieves the active route rule, protocol handler mode, and discovery metadata for a specific host.
  • POST /v1/config/apimap/promote: Promotes discovered or edited route rules live, atomically incrementing the configuration generation header X-Caber-Config-Version.
  • GET /v1/config/protocol-catalog: Returns the complete catalog of preconfigured protocol profiles defined in GLOBAL.protocolCatalog.
  • GET /v1/audit/traces: Returns paginated governance traces across csi_agg_events.
  • GET /v1/audit/traces/{trace_id}: Retrieves complete multi hop execution graphs, action waterfalls, and Verdict Snapshot v3 references for a specific trace ID.

4. OpenSearch Audit and Service Log Query Reference ​

Querying Enforced and Observed Redactions (csi_agg_events) ​

json
GET /csi_agg_events/_search
{
  "size": 25,
  "sort": [{ "event.time": { "order": "desc" } }],
  "query": {
    "bool": {
      "must": [
        { "terms": { "audit.governance_verdict": ["deny", "transform", "clarify"] } }
      ],
      "filter": [
        { "range": { "event.time": { "gte": "now-24h" } } }
      ]
    }
  },
  "_source": [
    "event.id",
    "event.time",
    "trace.trace_id",
    "user.name",
    "user.iss",
    "user.sub",
    "api.host.name",
    "api.path",
    "api.resource_type",
    "audit.was_redacted",
    "audit.governance_verdict",
    "audit.edge_verdict_snapshot_state",
    "audit.edge_verdict_snapshot_ref"
  ]
}

Querying Operator Service Log Errors and Recoveries (csi_service_log) ​

json
GET /csi_service_log/_search
{
  "size": 50,
  "sort": [{ "timestamp": { "order": "desc" } }],
  "query": {
    "bool": {
      "should": [
        { "terms": { "category": ["error", "alert"] } },
        { "wildcard": { "event": "*_recovered" } }
      ],
      "minimum_should_match": 1
    }
  }
}