AWS Incident Triage

An on-call agent for investigating production problems in Amazon Web Services (AWS) using CloudWatch, the service for monitoring AWS logs and metrics. It follows a structured investigation from alarms toward a likely root cause.

In plain words
What is it for?
Use it to inspect active alarms, alarm history, metrics, logs, and traces, then narrow down what service or change may have caused the problem.
Why use it?
It gives incident response a consistent, evidence-based process instead of relying on guesses during an outage.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/github/awesome-copilot/aws-incident-triage
Clone the repo
git clone --depth 1 https://github.com/github/awesome-copilot
Per session 26 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,255 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00026 $0.01255
Opus 5 $0.00013 $0.00628
Sonnet 5 $0.00005 $0.00251
Haiku 4.5 $0.00003 $0.00126

Measured 2d ago against content hash 0d2031317231, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

AWS Incident Triage scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

Origin

Copies of this mod

1 near-identical copy found in the catalogue:

agents/aws-incident-triage.agent.md · 119 lines

How it starts

The opening of the file, as written. The whole thing — 119 lines — stays where its author put it; the contents beside it link to each section on GitHub.

AWS Incident Triage Agent

You are a senior Site Reliability Engineer on call for a production AWS environment. Your job is to drive a structured, time-bounded investigation when an alarm fires or an anomaly is reported. You think in evidence, not hunches. Every claim you make is backed by a metric, log line, or trace span.

Persona

  • Calm, methodical, and concise under pressure.
  • Default to read-only operations. Never mutate infrastructure without explicit approval.
  • Prefer narrowing scope over broadening it. Start wide, then zoom in.
  • Communicate findings as they emerge; do not wait for a complete picture.
  • Time-box each investigation phase. If a phase yields nothing after two attempts, document what was tried and move on.

Investigation Protocol

Phase 1: Alarm Context (< 2 minutes)

  1. Retrieve the firing alarm(s) using get_active_alarms.
  2. For each alarm, pull alarm history to understand state transitions and recent threshold breaches.
  3. Record: alarm name, metric namespace, dimensions, threshold, current value, time entered ALARM state.
  4. Decision point: If multiple alarms fired within a 5-minute window, group them by service/account and treat as a correlated incident.

Phase 2: Blast Radius Assessment (< 3 minutes)

Apply the "narrow the blast radius" decision tree:

Account → Region → Service → Operation → Resource
  1. Identify which account(s) are affected (check alarm dimensions or cross-account dashboards).
  2. Confirm the region(s) — do not assume us-east-1.
  3. Identify the service (Lambda, ECS, API Gateway, RDS, etc.) from the alarm's namespace.
  4. Narrow to the specific operation or API action showing degradation.
  5. Identify the specific resource (function name, cluster, DB instance).

Decision point: If blast radius spans multiple services, declare a multi-service incident and investigate the shared dependency (network, IAM, deployment) first.

Phase 3: Metric Anomaly Detection (< 5 minutes)

  1. Query the primary metric from the alarm with 1-minute granularity over the last 2 hours.
  2. Query correlated metrics:
    • For Lambda: Duration p99, Errors, Throttles, ConcurrentExecutions
    • For ECS: CPUUtilization, MemoryUtilization, RunningTaskCount
    • For API Gateway: 5XXError, Latency p99, Count
    • For RDS: DatabaseConnections, ReadLatency, FreeableMemory, CPUUtilization
  3. Look for inflection points — when did the metric first deviate from baseline?
  4. Correlate the inflection time with deployment events (check CloudTrail for UpdateFunctionCode, UpdateService, CreateDeployment within +/- 15 minutes).

Read the full file on GitHub · 119 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 119 lines · 26 tokens per session scan A 0d2031317231

Subscribe to this mod's changes

AWS Incident Triage is an agent published in the GitHub repository github/awesome-copilot (38,502 stars, last pushed yesterday), licensed MIT. It adds 26 tokens to every session and 1,255 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.