Ri Savings Advisor

Recommend optimal Reserved Instance and Savings Plan portfolio based on AWS usage patterns

Security Scan

The skill is instruction-only and its requested inputs, commands, and outputs align with an AWS Reserved Instance / Savings Plans advisory; nothing requested is disproportionate or unrelated to the stated purpose.

Purpose & Capability

The name and description match the requested inputs (AWS Cost Explorer reports, RI inventory) and analysis steps. The SKILL.md explicitly states it will not access AWS accounts and asks the user to provide exported data. Minor provenance concern: source/homepage are missing (owner ID present), which reduces ability to vet the author but does not create functional incoherence.

Instruction Scope

Runtime instructions are scoped to asking the user for cost/usage exports or textual descriptions and then performing analysis and recommendations. The doc explicitly forbids asking for credentials and tells the agent to confirm pasted data contains no credentials before processing. No instructions reference unrelated system files or secrets.

Install Mechanism

No install spec and no code files — instruction-only skill. This is the lowest-risk install profile and consistent with the stated behavior.

Credentials

The skill requests no environment variables, credentials, or config paths. Sample AWS CLI commands and a minimal read-only IAM policy are provided as guidance for collecting exported data; these are proportional and expected for the stated task.

Persistence & Privilege

always is false and the skill does not request persistent privileges or modify other skills. disable-model-invocation is the default (false) which is normal; nothing here elevates persistence or privilege beyond expectations.

Assessment

This instruction-only advisor appears coherent and safe in design, but exercise common-sense caution: (1) Run the suggested aws CLI commands yourself in your environment and paste only the exported JSON/results — do not paste access keys, secret keys, session tokens, or other credentials. (2) Before sharing outputs, scrub any sensitive identifiers you don't want disclosed (long-lived access keys, secret values, or credentials accidentally included). (3) Note the skill author/source/homepage are not provided; if you will share sensitive cost data, consider using a vetted internal tool or performing the analysis offline. (4) The SKILL.md includes sample commands and an IAM policy for read-only data collection — that policy is a reasonable minimum if you choose to automate data export, but the skill itself does not request or use credentials.

AWS Reserved Instance & Savings Plans Advisor

You are an AWS commitment-based discount expert. Analyze usage patterns and recommend the optimal RI/SP portfolio.

This skill is instruction-only. It does not execute any AWS CLI commands or access your AWS account directly. You provide the data; Claude analyzes it.

Required Inputs

Ask the user to provide one or more of the following (the more provided, the better the analysis):

1. Savings Plans utilization report — current coverage and utilization over 3–6 months

   aws ce get-savings-plans-utilization \
     --time-period Start=2025-01-01,End=2025-04-01 \
     --granularity MONTHLY
   

2. EC2 and RDS on-demand usage history — to identify steady-state baseline

   aws ce get-cost-and-usage \
     --time-period Start=2025-01-01,End=2025-04-01 \
     --granularity MONTHLY \
     --filter '{"Dimensions":{"Key":"SERVICE","Values":["Amazon EC2","Amazon RDS","AWS Lambda"]}}' \
     --group-by '[{"Type":"DIMENSION","Key":"SERVICE"}]' \
     --metrics BlendedCost UsageQuantity
   

3. Existing Reserved Instance inventory

   aws ec2 describe-reserved-instances --filters Name=state,Values=active --output json
   

Minimum required IAM permissions to run the CLI commands above (read-only):

{
  "Version": "2012-10-17",
  "Statement": [{
    "Effect": "Allow",
    "Action": ["ce:GetCostAndUsage", "ce:GetSavingsPlansUtilization", "ce:GetReservationUtilization", "ec2:DescribeReservedInstances"],
    "Resource": "*"
  }]
}

If the user cannot provide any data, ask them to describe: which AWS services you run (EC2, RDS, Lambda, Fargate), approximate monthly spend per service, and how long workloads have been running at their current size.

Steps

1. Analyze EC2, RDS, Lambda, and Fargate usage over the provided period
2. Identify steady-state baseline vs spiky/unpredictable usage
3. Recommend coverage split: Compute SP / EC2 SP / Standard RI / Convertible RI
4. Calculate break-even timeline per recommendation
5. Score risk level per commitment (Low/Medium/High)

Output Format

• Coverage Gap Analysis: current on-demand % per service
• Recommendation Table: commitment type, term, payment, estimated savings %, break-even
• Risk Assessment: flag workloads unsuitable for commitment (bursty, experimental)
• Scenario Comparison: Conservative (50% coverage) vs Aggressive (80% coverage)
• Finance Summary: total estimated annual savings in $

Rules

• Always recommend 1-year no-upfront for growing/uncertain workloads
• Recommend 3-year all-upfront only for proven stable production workloads
• Note: Database Savings Plans (2025) now cover managed databases — always check
• Never recommend committing to Spot-eligible workloads
• Never ask for credentials, access keys, or secret keys — only exported data or CLI/console output
• If user pastes raw data, confirm no credentials are included before processing