This appendix provides checklists for maintainer reviews, autonomous remediation, rollback, runbook review, and operator handoffs.
Quarterly maintainer review
View diagram source
flowchart LR
Inputs[Signals and metrics] --> Review[Quarterly review]
Review --> Decisions[Decisions]
Review --> Backlog[Backlog]
Review --> Retire[Retirements]
Review --> Risks[Risks]Checklist:
- Review current-state references and mark stale claims.
- Review provider scorecards and trust tiers.
- Re-run critical contract tests.
- Review unresolved postmortem actions.
- Identify stale runbooks and schedule drills.
- Review cost per validated result by workload class.
- Review data retention and legal-hold exceptions.
- Identify adapters, APIs, or provider features to retire.
- Update training scenarios from recent incidents.
Autonomous action envelope
action_envelope:
agent_id: qops-remediator-prod
allowed_actions:
- annotate_target
- remove_target_from_routing_pool
- pause_low_priority_workloads
forbidden_actions:
- modify_pulse_parameters
- approve_public_claim
- delete_evidence
max_cost_impact_usd: 500
rate_limit: 5_actions_per_hour
approval_thresholds:
claim_grade_workload: human_required
physical_control_change: human_required
rollback_required: true
audit_stream: qops.agent.actionsView diagram source
flowchart TB
Agent[Agent] --> Envelope[Action envelope]
Envelope --> Policy[Policy evaluation]
Policy --> Action[Allowed action]
Action --> Verification[Verification]
Verification --> Record[Action record]Safe rollback checklist
Before action:
- Capture pre-state.
- Identify affected workloads and claims.
- Confirm rollback owner.
- Confirm rollback target state.
- Confirm validation canary.
- Confirm communication channel.
After action:
- Run validation canary.
- Compare post-state to expected envelope.
- Preserve logs and evidence.
- Close or escalate action record.
- Update runbook if actual steps differed.
View diagram source
stateDiagram-v2
[*] --> PreStateCaptured
PreStateCaptured --> ActionTaken
ActionTaken --> Validated
Validated --> Closed: pass
Validated --> Rollback: fail
Rollback --> Revalidated
Revalidated --> ClosedRunbook review
- Is the owner current?
- Is the trigger condition specific?
- Are commands, dashboards, and provider links current?
- Does the runbook preserve evidence before mutation?
- Does it include rollback and escalation?
- Has it been drilled in the last quarter?
- Is the runbook short enough for incident use?
View diagram source
flowchart LR
Runbook[Runbook] --> Drill[Drill]
Drill --> Findings[Findings]
Findings --> Patch[Patch]
Patch --> Approve[Approve]
Approve --> Active[Active]Operator handoff template
# Quantum operations handoff
- Time:
- Outgoing operator:
- Incoming operator:
- Active incidents:
- Affected targets:
- Workload freezes:
- Calibration or compiler changes in progress:
- Evidence gaps:
- Vendor tickets:
- Next decision deadline:
- Risks to watch:
- Confirmation from incoming operator:Legal hold activation
View diagram source
flowchart TB
Scope[Legal scope] --> IDs[Operational IDs]
IDs --> Freeze[Freeze records]
Freeze --> Provider[Provider preservation request]
Freeze --> Validate[Validate coverage]
Validate --> Report[Hold status report]Checklist:
- Confirm scope with legal.
- Convert scope to workload, claim, incident, provider job, and time identifiers.
- Lock internal evidence packages.
- Freeze deletion policies for matching records.
- Notify provider if preservation is contractually available.
- Record acknowledgement and exceptions.
- Validate hold coverage with sample queries.
Maintainer dashboard fields
| Metric | Review question |
|---|---|
| stale references | Which current-state claims need source review? |
| failing contract tests | Which interfaces or semantics drifted? |
| orphan evidence | Which results lack complete lineage? |
| unresolved RCA actions | Which incidents did not produce durable learning? |
| scorecard regressions | Which providers or targets lost trust? |
| runbook exercise age | Which procedures are untested? |