Skip to main content

​
Overview

Annie Instructions let you add guidance on top of Annie’s memory. Annie already knows your infrastructure from connected integrations and learned patterns—instructions tell her how to respond in specific situations. Use instructions to control formatting, prioritize certain checks, or customize behavior for different alert types.

​
Instruction Types

Set baseline instructions that Annie follows for all responses:
FieldDescriptionExample
InstructionsWhat Annie should do”Always check CloudWatch logs first”
FormattingHow Annie structures responses”Use bullet points, include severity”
Create rules that trigger based on specific conditions. When the trigger condition matches an alert, Annie follows those specific instructions.
FieldDescriptionExample
Trigger ConditionWhen this instruction applies”Alert contains ‘database‘“
InstructionsWhat Annie should do”Check RDS metrics and recent schema changes”
FormattingResponse structure”Include connection pool stats”

​
Examples

Trigger: Alert contains "RDS" or "database" or "postgres" or "mysql"

Instructions:
- Check recent schema migrations in the last 24h
- Review connection pool metrics and active connections
- Look for long-running queries (>30s)
- Check for recent deployments that might have changed queries

Formatting:
- Include query execution times
- Show connection pool utilization percentage
- List any blocked queries

Trigger: Severity is "critical" or "P1" or "SEV1"

Instructions:
- Start with a 2-sentence executive summary
- Identify blast radius immediately
- Check all dependent services for cascading failures
- Prepare rollback options if a recent deployment is involved
- Tag @oncall-team in the response

Formatting:
- Lead with executive summary
- Use severity labels (🔴 Critical, 🟡 Warning)
- Include estimated time to resolution if possible

Trigger: Alert from "AWS Budgets" or "cost anomaly" or "billing"

Instructions:
- Identify top 3 cost drivers for the alert period
- Check for runaway resources (forgotten instances, orphaned volumes)
- Compare current spend with last week's baseline
- Look for recent scaling events or new resource provisioning

Formatting:
- Show cost comparison table (current vs baseline)
- Highlight percentage increase
- List specific resources causing the spike

Trigger: Alert contains "pod" or "kubernetes" or "k8s" or "container"

Instructions:
- Check pod events for OOMKilled, CrashLoopBackOff, ImagePullBackOff
- Review resource limits vs actual usage
- Check recent deployments and rollouts
- Look for node pressure issues

Formatting:
- Include pod status and restart count
- Show resource usage (CPU/memory) vs limits
- List recent events chronologically

​
Get Started

Create Account

Sign up for Anyshift

Request Demo

See Annie Instructions in action