Current control
Loading active collaboration state.
The current incident story will appear here once the workspace loads.
Waiting for incident selection.
Next operator action will appear here once the incident loads.
SENTINEL handed evidence to PRISM, PRISM synthesized the diagnosis, FORGE prepared the runbook, and GUARDIAN now controls the final safety gate.
SENTINEL classifies → PRISM diagnoses → REPLICA reproduces → TRACE debugs → FORGE recommends → GUARDIAN approves
The selected input bundle proof will appear here for fresh incidents created from /inputs.
The operator guidance from the chosen bundle will appear here.
NEXUS will separate extracted evidence from inferred reasoning here for fresh incidents.
Operator guidance for fresh intake will appear here.
Decision pending.
Guardian is the explicit safety gate. Execution stays subordinate to bot collaboration until this check is clear.
The runbook summary will appear here so the operator understands what approval would actually execute.
Execution impact and rollback posture will appear here once the incident loads.
The current incident story will appear here once the workspace loads.
Waiting for evidence, prior incidents, and policy checks.
Waiting for the crew to reach a governed decision.
Waiting for incident context.
Waiting for correlated evidence.
Waiting to produce a rollback-ready proposal.
Waiting for the runbook to clear safety review.
LangGraph orchestrates the crew, so the diagnosis can loop back to earlier evidence if a later branch changes the confidence story.
Default mode stays deterministic. Add your own key only when you want live FORGE reasoning for this session.
No user key attached. The app is in deterministic demo mode.
Classification bot
Waiting for incident context.
Raw input becomes an evidence bundle.
Diagnosis bot
Waiting for correlated evidence.
Evidence becomes a diagnosis packet.
Reproduction bot
Waiting to validate in sandbox.
Diagnosis becomes a reproduction packet.
Debugging bot
Waiting to inspect code paths.
Reproduction becomes a debug packet.
Remediation bot
Waiting to shape the safest runbook.
Diagnosis becomes a runbook proposal.
Governance Bot
Waiting to approve or block execution.
Runbook proposal becomes a governance decision.
Waiting for FORGE to prepare the operator-facing action.
Waiting for REPLICA to confirm whether runtime evidence is available.
Waiting for TRACE to narrow the first debugging boundary.
Loading prior incidents, runbook history, and unresolved work.
The runtime stays deterministic and bounded even when one branch needs a fallback.
Reproduction findings will appear here once the investigation packet loads.
Runtime host posture will appear here once REPLICA loads.
Runtime mode details will appear here once REPLICA loads.
Replay capability details will appear here once REPLICA loads.
Replay trigger status will appear here once REPLICA loads.
Replay lifecycle state will appear here once REPLICA loads.
Replay trust packet will appear here once REPLICA loads.
Runtime comparison details will appear once REPLICA validates a mitigation.
Debugging hints will appear here once TRACE narrows a likely path.
Replay evidence will appear here once TRACE has a runtime-backed clue set.
Inspect-here-first guidance will appear here once TRACE narrows a likely path.
Developer handoff details will appear here once TRACE prepares a bounded packet.
Risk framing will appear here once TRACE prepares guidance.
Scope boundaries will appear here once TRACE prepares guidance.
Confidence caveats will appear here once TRACE prepares guidance.
Next steps if divergent will appear here once TRACE prepares guidance.
No raw incident text captured yet.
This timeline will show the source channel, authentication, enrichment, evidence, agents, safety, and outcome steps.
Audit trail size will appear here once the incident loads.
Execution remains behind an explicit Guardian approval or block choice.
Track when and where the investigation packet was sent downstream.
Track engineering team feedback and outcome confirmation.