The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The preflight verification script recommends downloading the uv package manager directly from the official vendor's site.
Evidence: curl -LsSf https://astral.sh/uv/install.sh | sh suggested in scripts/preflight_check.py.
Context: The download targets the official domain of a well-known development tool provider and is documented as a prerequisite for Python-based agent runtimes.
[DATA_EXFILTRATION]: The configuration scanning tool includes logic to prevent the accidental exposure of secrets during the migration process.
Evidence: SECRET_PATTERNS regex in scan_configs.py identifies keys, tokens, and passwords for redaction.
Impact: Sensitive values are replaced with <REDACTED> in the migration inventory, and users are guided toward secure AWS Secrets Manager integration via the agentcore identity credential provider.
[COMMAND_EXECUTION]: Automation scripts execute shell commands to facilitate environment setup and project configuration.
Evidence: Usage of subprocess.run() in preflight_check.py and generate_project.py.
Operations: Commands are used to verify Node.js and Python versions and to call aws sts get-caller-identity to populate configuration files with the correct AWS account ID.
[REMOTE_CODE_EXECUTION]: The skill automates the generation of agent execution code and container specifications for deployment to AWS.
Evidence: Templates in references/templates/ and generated code in assets/poc-cve-verify/ (e.g., main.py).
Safety: Generated code follows static entrypoint patterns using the bedrock-agentcore library and does not employ dynamic evaluation of untrusted input.

agentcore