Scenario_id stringlengths 14 23 ⌀ | task_type stringclasses 6
values | incident_input stringlengths 137 288 ⌀ | memory_table stringlengths 382 858 ⌀ | available_tools stringlengths 35 124 ⌀ | agent_plan stringlengths 85 265 ⌀ | expected_agent_actions stringlengths 53 144 ⌀ | expected_final_output stringlengths 101 253 ⌀ | evaluation_criteria stringlengths 57 175 ⌀ |
|---|---|---|---|---|---|---|---|---|
null | null | null | null | null | null | null | null | null |
cost_spike_001 | Cost Ops | Slack
We’re seeing a ~45% increase in our AWS bill for December compared to November.
EC2 usage looks much higher, RDS costs nearly doubled, and S3 increased slightly.
No major launches or experiments were planned this month.
Please investigate and share findings with recommendations.
| Date | Service | Region | Resource_Type | Usage | Cost($)
-----------------------------------------------------------------------
2025-11-01 | EC2 | us-east-1 | t3.large | 420 hrs | 840
2025-11-05 | EC2 | us-east-1 | t3.large | 480 hrs | 960
2025-11-10 | RDS | us-eas... | get_cost_and_usage_by_service
get_monthly_cost_breakdown
detect_cost_anomalies
list_idle_ec2_instances
list_ebs_volumes | Retrieve cost data for November and December grouped by service
Detect anomalies exceeding 20% month-over-month increase
Correlate EC2 cost increase with instance uptime
Identify idle EC2 instances and unused storage
Summarize findings and propose optimizations | Call cost comparison APIs
Run anomaly detection on EC2, RDS, S3
Query idle resources contributing to cost
Generate a structured Slack summary | EC2 and RDS are primary cost drivers
EC2 usage increased ~60% due to long-running instances
RDS usage doubled due to extended uptime
S3 growth is within expected variance
Recommend stopping idle EC2s, deleting orphaned EBS, and reviewing RDS scaling | Correct identification of cost drivers
Accurate baseline comparison
Evidence-backed reasoning
Actionable cost-saving recommendations
Clear, professional Slack-style report |
perf_degradation_002 | Reliability | (Teams – #platform-alerts)
Over the last 3 weeks, p95 latency for the Orders API
has steadily increased from ~250ms to ~900ms.
No major releases were deployed during this time.
Customer complaints are increasing. | Date | Service | Metric | Value
---------------------------------------------
2025-11-05 | OrdersAPI| p95 latency | 250ms
2025-11-10 | OrdersAPI| p95 latency | 310ms
2025-11-15 | OrdersAPI| p95 latency | 420ms
2025-11-20 | OrdersAPI| p95 latency | 510ms
2025-11-25 | OrdersAPI| p95 latency ... | get_service_metrics
compare_historical_performance
detect_dependency_latency | Compare latency trends over time
Check for gradual degradation patterns
Identify dependency or resource saturation
Propose reliability improvements | Pull historical latency metrics
Correlate with downstream services
Identify infra or config drift | Gradual degradation indicates non-code-related issue
Database connection pool saturation detected
Recommend pool tuning and latency SLO alerts | Recognizes gradual vs sudden failure
Correct root cause hypothesis
Reliability-focused mitigation |
security_alert_003 | Security | (Email – Security Operations)
We detected a spike in failed IAM login attempts
from unfamiliar IP ranges over the last 48 hours.
No confirmed breach yet.
Please investigate and advise on next steps.
| Date | Source IP | Event | Count
------------------------------------------------------
2025-12-10 | 103.21.x.x | Failed IAM Login | 120
2025-12-10 | 91.142.x.x | Failed IAM Login | 95
2025-12-11 | 103.21.x.x | Failed IAM Login | 180
2025-12-11 | 91.142.x.x | Failed IAM Login ... | get_auth_logs
list_anomalous_ips
enforce_mfa | Analyze login failure patterns
Identify anomalous IP behavior
Assess IAM hardening gaps | Query auth logs
Detect brute-force patterns
Recommend preventive controls | Coordinated brute-force attempt detected
MFA missing on targeted IAM users
Recommend MFA enforcement and IP blocking | Correct threat classification
No false breach claims
Preventive security posture |
resource_leak_004 | Ops Excellence | (Slack – #sre)
Memory usage on the Payments service pods
keeps increasing over time.
Pods restart every 30–40 hours due to OOM kills.
Traffic levels are stable.
| Date | Pod Name | Memory Used
----------------------------------
2025-12-01 | pay-a12 | 1.1 GB
2025-12-02 | pay-a12 | 1.3 GB
2025-12-03 | pay-a12 | 1.5 GB
2025-12-04 | pay-a12 | 1.7 GB
2025-12-05 | pay-a12 | 1.9 GB
2025-12-06 | pay-a12 | 2.1 GB
2025-12-07 | pay-a12 | 2.3 GB
2025-12-08 | pay-a12 | 2.5 GB
... | get_pod_metrics
analyze_oom_events | Examine memory growth trend
Rule out traffic-based scaling
Identify potential leaks | Pull pod-level memory metrics
Analyze restart and OOM patterns | Clear memory leak pattern detected
Recommend heap dump and profiling
Temporary memory limit increase for stability | Correct leak identification
Balances short-term and long-term fixes |
backup_failure_005 | Reliability | (MoM – Weekly Infra Review)
Daily RDS automated backups for the production database
failed for multiple consecutive days.
No alerts were triggered.
| Date | Backup Type | Status
--------------------------------
2025-12-01 | Automated | Success
2025-12-02 | Automated | Success
2025-12-03 | Automated | Failed
2025-12-04 | Automated | Failed
2025-12-05 | Automated | Failed
2025-12-06 | Automated | Failed
2025-12-07 | Automated | Failed
2025-12-08 ... | check_backup_jobs
validate_alerting_rules | Investigate backup job failures
Identify alerting gaps
Recommend reliability improvements | Query backup status logs
Review alert configurations | Backup misconfiguration identified
Alerting failure is a secondary risk
Recommend backup validation and alerts | Data durability awareness
Identifies silent failure risks |
network_latency_006 | Workload Assessment | (Teams – #network-ops)
Users from APAC regions are reporting slow load times
on the analytics dashboard since last week.
US and EU users report normal performance.
No recent infra changes were announced.
| Date | Region | Avg Latency | p95 Latency
--------------------------------------------------
2025-12-01 | us-east-1 | 130ms | 210ms
2025-12-01 | ap-south-1| 480ms | 720ms
2025-12-03 | ap-south-1| 520ms | 780ms
2025-12-05 | ap-south-1| 560ms | 820ms
2025-12-06 | ap-south-1| 590ms ... | get_network_latency_metrics
analyze_traffic_routing
check_regional_endpoints | Compare latency across regions
Identify cross-region traffic routing
Assess workload placement | Retrieve region-wise latency metrics
Validate endpoint routing configuration | APAC traffic routed to us-east-1
No regional endpoint deployed
Recommend APAC deployment or edge caching | Region-aware analysis
Correct workload placement diagnosis
Practical performance optimization |
high_memory_007 | App Modernization | (Email – Engineering Leadership)
After upgrading the JVM, the legacy reporting service
now consumes significantly more memory.
Memory usage is 4x higher than last quarter.
We need guidance on whether to tune or modernize.
| Date | Service | Avg Memory Usage
-----------------------------------------------
2025-09-01 | ReportingSvc | 1.2 GB
2025-09-15 | ReportingSvc | 1.4 GB
2025-10-01 | ReportingSvc | 1.6 GB
2025-10-15 | ReportingSvc | 1.9 GB
2025-11-01 | ReportingSvc | 2.3 GB
2025-11-15 | Reportin... | get_jvm_metrics
analyze_gc_behavior
check_service_architecture | Analyze memory growth trend
Evaluate GC and heap efficiency
Assess modernization feasibility | Pull JVM heap and GC metrics
Compare tuning vs redesign tradeoffs | Legacy architecture poorly suited for current JVM
GC tuning provides limited relief
Recommend container-native refactor or service split | Strategic thinking beyond tuning
Clear modernization rationale
Balanced tradeoff analysis |
unexpected_shutdown_008 | Reliability | (Slack – #incident-response)
Production application node shut down unexpectedly
at 02:14 AM IST.
No deployments or scaling events were recorded.
Investigation needed urgently.
| Date | Metric | Value
---------------------------------
2025-12-10 | Disk Usage | 82%
2025-12-11 | Disk Usage | 86%
2025-12-12 | Disk Usage | 90%
2025-12-13 | Disk Usage | 94%
2025-12-13 | Disk Usage | 96%
2025-12-14 | Disk Usage | 98%
2025-12-14 | CPU Usage | Normal
2025-12-... | get_node_metrics
analyze_system_logs
check_disk_usage | Examine resource utilization prior to shutdown
Identify triggering threshold
Recommend preventive measures | Retrieve disk and system logs
Correlate shutdown with resource exhaustion | Disk exhaustion caused node shutdown
No disk utilization alerts configured
Recommend alerts and automated cleanup | Accurate root cause identification
Preventive reliability mindset |
api_error_spike_009 | Ops Excellence | (Teams – #app-support)
Public API error rate increased from ~0.3% to 6%
within 20 minutes.
Multiple customers report failed requests.
| Date | API Endpoint | 5xx Error Rate
-----------------------------------------
2025-12-14 | /checkout | 0.3%
2025-12-14 | /checkout | 0.4%
2025-12-14 | /checkout | 1.2%
2025-12-14 | /checkout | 2.8%
2025-12-14 | /checkout | 4.1%
2025-12-14 | /checkout | 5.3%
2025-12-14 | /checkout | 6... | get_api_metrics
trace_dependency_failures
inspect_timeout_configs | Analyze error spike timeline
Correlate with downstream dependencies
Propose resilience improvements | Pull API and dependency metrics
Identify cascading failure patterns | Downstream payment service latency caused failures
No circuit breaker or retry policy
Recommend resilience patterns and rate limiting | Correct failure cascade reasoning
Ops maturity awareness |
storage_quota_010 | Cost Ops | (Email – Data Platform Team)
Our primary S3 bucket has reached 90% of its quota.
Unexpected growth observed in logs/ prefix.
We risk hitting hard limits soon.
| Date | Prefix | Storage Used
------------------------------------
2025-11-01 | logs/ | 6.2 TB
2025-11-05 | logs/ | 6.8 TB
2025-11-10 | logs/ | 7.4 TB
2025-11-15 | logs/ | 8.1 TB
2025-11-20 | logs/ | 8.9 TB
2025-11-25 | logs/ | 9.6 TB
2025-12-01 | logs/ | 10.4 TB
2025-12-05 ... | analyze_storage_usage
list_lifecycle_policies
estimate_storage_costs | Analyze storage growth trends
Identify misconfigured retention
Recommend cost controls | Query prefix-level storage usage
Inspect lifecycle policy configuration | Log retention misconfigured
No lifecycle policies applied
Recommend automated archival and deletion | Correct cost-growth attribution
Preventive cost governance |
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
| Field | Description |
|---|---|
Scenario_id |
Unique ID |
task_type |
Category of the task |
context |
Problem or incident narrative |
memory_table |
Structured information /logs |
available_tools |
tools that the agent can use |
agent_plan |
Step-by-step plan |
expected_agent_actions |
Sequence of actions with tool calls |
expected_final_output |
Final answer results |
evaluation_criteria |
Metrics/rubric for correctness |
- Downloads last month
- 5