The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes various AWS CLI commands (e.g., aws ec2 describe-instances, aws cloudwatch get-metric-statistics, aws ce get-cost-and-usage) to inspect infrastructure and verify costs. These operations are limited to read-only resource discovery and metrics collection, which is necessary for identifying optimization targets.
[EXTERNAL_DOWNLOADS]: The skill uses fetch to retrieve cost optimization guidelines from official AWS documentation (docs.aws.amazon.com). This is a trusted source used solely to provide reference material for the analysis.
[DATA_EXFILTRATION]: The tool processes sensitive configuration and cost data from the user's AWS account. This data is used to generate recommendations that are subsequently posted as issues in a user-identified GitHub repository. The information transfer is transparent and follows the primary purpose of the skill.
[PROMPT_INJECTION]: The skill processes untrusted external content and local files, creating a surface for indirect prompt injection.
Ingestion points: Fetched AWS documentation and local IaC files (.tf, .yaml, .ts, etc.) as defined in SKILL.md.
Boundary markers: No specific delimiters are defined to separate untrusted content from agent instructions.
Capability inventory: AWS CLI execution, local file system read access, and GitHub issue creation via an MCP server.
Sanitization: No explicit validation or filtering is performed on the ingested documentation or configuration data.

aws-cost-optimize