The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently executes system-level commands to perform its migration tasks. It uses shell commands to inspect billing exports for headers, runs the terraform CLI for formatting and validating generated infrastructure code, and executes dynamically generated Python or JavaScript scripts to process local billing data. These operations are required for the tool's primary discovery and validation functions.- [CREDENTIALS_UNSAFE]: The tool handles highly sensitive data as part of its secret migration workflow. It generates shell scripts (e.g., 04-migrate-secrets.sh) that use the gcloud and aws CLIs to extract secrets from GCP Secret Manager and upload them to AWS Secrets Manager. This process involves handling cleartext sensitive information through local temporary files and command-line arguments.- [REMOTE_CODE_EXECUTION]: During the generation phase, the skill executes terraform init to validate the created infrastructure configurations. This command automatically downloads and executes provider plugins from external registries. While these are typically from well-known sources like HashiCorp or AWS, it constitutes the execution of remote binaries on the host system.- [EXTERNAL_DOWNLOADS]: The skill generates deployment and setup scripts that instruct the user to install various dependencies, such as bedrock-agentcore-cli, strands-agents, and litellm, via package managers like pip. It also depends on the awspricing MCP server to fetch real-time pricing data.- [DATA_EXFILTRATION]: The skill includes an optional feedback phase that builds an anonymized telemetry trace (trace.json) containing migration metadata such as resource counts, spend, and mapped services. The skill provides the user with a single-line minified version of this data and directs them to submit it to a survey domain owned by a well-known service provider (pulse.amazon).

gcp-to-aws