TrustAgents is a defensive-security research repository for simulated fraud detection in compliance evidence pipelines. It supports proposal review with synthetic artifacts and explicit limits. TrustSignal production controls remain unchanged: artifact hashing, signed receipt issuance, receipt verification, scoped auth, webhook integrity checks, and redaction or partner-gated surfaces are not replaced by this repository.
The trustagents oracle service in src/trustagents/ is:
- experimental
- optional
- non-blocking
- isolated from the production verification dependency chain
Oracle output is advisory and ingestion-ready only. Current repository content does not claim production dependency on oracle fusion, chain anchoring, ZKP attestation, or autonomous research agents. Any future integration is gated and documented as future work.
All fixtures and examples are synthetic unless explicitly documented otherwise.
proposal/: grant-ready summaries, application text, budget, and timelinedocs/: architecture boundaries, schema contracts, methodology, and threat modeldatasets/: synthetic schema and sample anomaliessimulations/: scenario catalog and pipeline modelagents/: role definitions for research agentseval/: benchmark design and metricssrc/trustagents/: isolated experimental oracle evaluation servicetests/: unit, integration, and regression checks for the oracle service