Skip to content

Evidence Export Guide

Phase 4 through Phase 6 converge on one export story:

trace -> decision -> signed evidence -> package

This page explains the export layers, the expected consumers, and the adjacent JSONL/SIEM/OpenInference paths that matter for enterprise workflows.

Export Layers

LayerProducerAudience
Operator evidence packagesf_operator.export_package()operators, incident responders, security reviewers
Enterprise evidence packagesf_enterprise.generate_evidence_package()auditors, compliance teams, architecture review boards
JSONL event exportJSONLExporter / spanforge.ioarchival, offline analysis, replay inputs
SIEM exportSplunk / syslog exportersSOC, SIEM, alerting, correlation systems
OpenInference bridgeOpenInferenceSpanBridgeinteroperability with OpenInference-style trace consumers

Operator Package

sf_operator.export_package(trace_id) packages one runtime trace into:

  • final outcome
  • concise why-allowed or why-blocked summary
  • policy decisions
  • explanations
  • grounding results
  • scope and RBAC decisions
  • lineage records
  • signed audit records
  • timeline view
  • chain-verification result

CLI:

spanforge operator inspect TRACE_ID
spanforge operator export TRACE_ID --output operator-package.json

Enterprise Package

sf_enterprise.generate_evidence_package(trace_id) wraps the operator package with enterprise context:

  • deployment profile
  • retention and export controls
  • enterprise status
  • reference deployment architectures
  • signed checksum and signature

Use this when the consumer needs to understand not only the trace evidence, but also how SpanForge is deployed and controlled in the target environment.

JSONL and Audit Workflows

JSONL remains the simplest durable transport for:

  • offline archives
  • replay inputs
  • air-gapped evidence movement
  • long-term audit retention

Relevant docs:

SIEM-Friendly Mapping

Phase 5 introduced a normalized SIEM mapping so Splunk, syslog, and other downstream systems see the same core event fields:

  • event identity and timestamps
  • trace/span linkage
  • actor and session metadata
  • normalized category
  • normalized severity
  • payload and tags

That normalized schema is used by the Splunk HEC and syslog/CEF exporters.

OpenInference Interop

The OpenInference bridge translates existing SpanForge spans into an OpenInference-style trace document. It is intended for interoperability, not for replacing SpanForge’s own signed evidence chain.

Use it when you need:

  • compatibility with OpenInference-aware tooling
  • a bridge from SpanForge traces into external analysis pipelines
  • a vendor-neutral trace handoff alongside the signed governance exports

Related Docs