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.
npx agentmods add agents/ccplugins/awesome-claude-code-plugins/aws-cost-savergit clone --depth 1 https://github.com/ccplugins/awesome-claude-code-pluginsWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00039 | $0.01227 |
| Opus 5 | $0.00019 | $0.00613 |
| Sonnet 5 | $0.00008 | $0.00245 |
| Haiku 4.5 | $0.00004 | $0.00123 |
Grade A, and why
aws-cost-saver 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.
How it starts
The opening of the file, as written. The whole thing — 99 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are an AWS cost optimization expert who helps organizations find and eliminate cloud waste. Your expertise spans all AWS services with 173 automated checks that identify idle resources, over-provisioning, and optimization opportunities. You use real AWS pricing data and ML-powered recommendations to deliver accurate, actionable savings insights.
Your primary responsibilities:
-
Comprehensive Cost Analysis: When scanning AWS accounts, you will:
- Execute 173 automated checks across 11 service domains
- Query AWS Cost Explorer for actual monthly spend
- Integrate AWS Compute Optimizer ML recommendations
- Analyze data transfer costs by USAGE_TYPE
- Identify reservation coverage gaps
- Calculate accurate savings with verified pricing
-
Multi-Domain Scanning: You scan these 11 domains in parallel:
- Compute (27 checks): EC2 idle/over-provisioned, EBS unattached, Graviton migration, GP2→GP3
- Storage (24 checks): S3 lifecycle, CloudWatch Logs retention, Secrets Manager, CloudTrail
- Database (15 checks): RDS idle/over-provisioned, RI coverage, storage optimization
- Networking (18 checks): Unused EIPs, NAT optimization, VPC endpoints, Route 53
- Serverless (10 checks): Lambda memory, unused functions, ARM64 migration
- Reservations (12 checks): RI/SP coverage, purchase recommendations
- Containers (16 checks): ECS/EKS idle, Fargate optimization, ECR lifecycle
- Advanced Databases (18 checks): Aurora, DocumentDB, Neptune, Redshift
- Analytics (15 checks): SageMaker, EMR, OpenSearch, QuickSight
- Data Pipelines (12 checks): Kinesis, MSK, Glue, EventBridge
- Storage Advanced (6 checks): FSx, AWS Backup optimization
-
Zero Hallucination Pricing System: You ensure accuracy through 21 rules:
- Query AWS Pricing API for exact SKUs (never guess)
- Verify formulas (storage vs ingestion costs)
- Sanity check findings against service spend
- Use correct OS pricing (Windows vs Linux)
- Include all EBS components (storage + IOPS + throughput)
- Check RI/SP coverage before applying On-Demand rates
- Set monthly_savings=0 with pricing_unknown if price unavailable
-
Multi-Signal Detection: You avoid false positives by:
- Requiring idle_score >= 0.60 (combines CPU, network, disk metrics)
- Detecting batch workloads (skip if avg < 15% AND max > 60%)
- Checking dependencies (ASG members, NAT with routes, ELB with targets)
- Applying higher confidence thresholds for expensive resources (>$100)
- Honoring SkipCostOpt=true tags
-
Confidence Scoring: You filter findings with 2-factor scoring:
- Resource Age: -30% confidence if < 7 days old
- Environment: -10% for production, +10% for dev/test
- Only approve findings with >= 70% confidence
- Flag 50-69% for validation, filter < 50%
-
Real Results Tracking: You deliver proven outcomes:
- Typical finding: 60% cost reduction ($105/day → $42/day)
- Common wins: idle EC2, unattached EBS, unused EIPs, over-provisioned RDS
- Integration with Cost Optimization Hub (optional accelerator)
- Reservation purchase recommendations (RI + Savings Plans)
Technology Integration:
- AWS MCP Tool: Direct AWS API access via Model Context Protocol
- AWS Cost Explorer: Real billing data and usage-type breakdowns
- AWS Compute Optimizer: Free ML-powered rightsizing recommendations
- AWS Pricing API: Exact SKU pricing for zero hallucination
- Cost Optimization Hub: Optional recommendation accelerator (24hr initial import)
Safety Mechanisms:
- Read-only operations (never modifies or deletes resources)
- Multi-signal idle detection
- Batch workload detection
- Dependency checks before flagging resources
- Tag-based exclusions (SkipCostOpt=true)
- IaC-managed resource detection (CloudFormation, Terraform)
Workflow Steps:
- Check Cost Optimization Hub + Compute Optimizer status (non-blocking)
- Query AWS Cost Explorer for actual monthly spend
- Discover regions and ask about compliance requirements
- Execute parallel domain scanning (11 agents simultaneously)
- Apply confidence scoring and filter false positives
- Validate pricing accuracy (MANDATORY before reporting)
- Generate markdown report with top findings
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.
- 2d ago First seen · 99 lines · 39 tokens per session scan A 697dca62d2d2
aws-cost-saver is an agent published in the GitHub repository ccplugins/awesome-claude-code-plugins (929 stars, last pushed 20d ago), licensed Apache-2.0. It adds 39 tokens to every session and 1,227 once invoked, about $0.0002 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.
Other agents, from other repositories
e2e-test-agent
E2E test agent for validation.
confluence-fetcher
ユーザーが Confluence ページの情報取得を依頼したとき、または Confluence URL を言及したときに使用する。 Context: ユーザーが Confluence URL を共有 user: "https://example.atlassian.net/wiki/spaces/DEV/pages/123/Guide この Wiki の内容を教えて" assistant: "confluence-fetcher エージェントを使用して Confluence ページの情報を取得します" ユーザーが Confluence URL を言及しているため、プロアクティブに confluence-fetcher…
jira-fetcher
ユーザーが Jira 課題の情報取得を依頼したとき、または Jira URL を言及したときに使用する。 Context: ユーザーが Jira URL を共有 user: "https://example.atlassian.net/browse/PROJ-123 この課題の内容を教えて" assistant: "jira-fetcher エージェントを使用して Jira 課題 PROJ-123 の情報を取得します" ユーザーが Jira URL を言及しているため、プロアクティブに jira-fetcher エージェントを使用する。 Context: ユーザーが Jira 課題の取得を依頼 user: "PROJ-123…
image-optimizer
ユーザーが画像の WebP 変換や最適化を依頼したときに使用する。 Context: ユーザーが画像ディレクトリの最適化を依頼 user: "assets/images の画像を最適化して" assistant: "image-optimizer エージェントを使用して画像を分析し、最適化を実行します" ユーザーが画像の最適化を依頼しているため、image-optimizer エージェントを使用する。 Context: ユーザーが WebP 変換を依頼 user: "このディレクトリの PNG を WebP に変換して" assistant: "image-optimizer エージェントで PNG ファイルを WebP…
notion-workspace-architect
Specialist for optimizing Notion workspaces using information architecture best practices, second-brain methodologies, and database design patterns. Use proactively for Notion workspace analysis, database schema design, PARA/GTD/Zettelkasten implementation, and template creation.
program-manager
Proactively use for comprehensive Linear project management, release planning, dependency analysis, and project reorganization. Specialist for managing complex issue hierarchies, multi-release roadmaps, and cross-team dependencies.