Expanded capabilities
- Compose control-posture assessments across prior Golden Path stages
- Highlight exposure / containment candidates for human review
- Read approval and NUNTIUS state without changing them
- Pin RECORD_ONLY findings when posture needs attention
- Stay observing-only on writes — densify/pin paths are operator-gated
How they learn & improve
VIGILIA improves which exposure keys matter from operator pins and FEEDBACK — it does not auto-contain or auto-escalate.
- Observe MONETA/ASSESS/NUNTIUS/approval posture with provenance.
- Surface containment candidates as RECORD_ONLY pins.
- Recalibrate attention weights only after human-applied learning.
How they work together
VIGILIA watches the control plane while HERMES watches flow health and TEMPUS proposes next human steps — none of them approve or send.
- HERMESPairs flow pressure with control exposure for operators.
- TEMPUSFeeds posture into short remediation plan blocks.
- Approval queueReads pending approvals as risk context only.
- JANUSAligns control posture with governance readiness cues.
Practical benefits
- See exposure before dispatchOperators spot unfinished approvals or risky stage combos before anything outbound runs.
- Densify on demandHuman densify actions deepen evidence without inventing incidents.
- Learning that stays reversibleWrong exposure keys become candidates you can reject — not a silent security autopilot.
Learning and improvement stay human-visible: candidates are RECORD_ONLY until an operator reviews and applies them. Sensitive sends, publishes, and approvals never run silently.
