📖 RunbooksAutomated Response

Automated incident response playbooks — detect, remediate, resolve.

📖
6
Active Runbooks
🔴
5
Armed
124
Executions (30d)
97.3%
Auto-Resolution

Cost Runaway ContainmentARMED

Throttle → model fallback → kill chain when agent cost exceeds budget threshold.

all_agents
1d ago
WHEN cost_per_hour > budget_limit * 0.8
1. throttle
2. switch_model
3. kill
47 runs2m 14s avg

Error Cascade BreakerARMED

Circuit-break → diagnostic eval → fallback model on sustained error spikes.

all_agents
1d ago
WHEN error_rate > 5% FOR 3 minutes
1. circuit_break
2. run_diagnostic
3. switch_model
23 runs3m 7s avg

Token Drift CorrectionARMED

Reinforce prompt constraints and cap max_tokens when output length drifts.

research-analyst
2d ago
WHEN token_output_drift > 25%
1. prompt_reinforce
2. set_param
3. run_eval
12 runs45s avg

Security QuarantineARMED

Immediate I/O block + evidence snapshot + mandatory human approval.

all_agents
3d ago
WHEN security_alert.severity == critical
1. block_io
2. snapshot_evidence
3. require_approval
3 runs7m avg

Latency SLA GuardianARMED

Identify bottleneck → model/tool/network-specific remediation.

all_agents
1d ago
WHEN p99_latency > slo_target * 1.5 FOR 5 minutes
1. identify_bottleneck
2. remediate
3. verify_slo
31 runs1m 36s avg

Budget Freeze ProtocolDISARMED

Freeze non-critical ops, queue deferred tasks, auto-resume at budget reset.

all_agents
5d ago
WHEN daily_spend >= daily_budget
1. freeze_non_critical
2. queue_tasks
3. schedule_resume
8 runs

Execution History

Select a runbook to view execution history.