The Agent Skills Directory

[Secure Data Handling]: The skill includes a mandatory confirmation prompt in the SKILL.md file that must be presented to the user before any local source code is uploaded. This ensures the user is aware of and consents to the analysis process. Additionally, the scripts/archive.py script implements a robust exclusion list to prevent the accidental upload of common sensitive files, such as .env files, SSH keys, and AWS credentials.
[Standard Authentication Practices]: The scripts/auth.py script manages IAM Identity Center authentication using the PKCE flow. It stores authentication tokens in the user's home directory (~/.launch-with-aws/session.json) using restricted file permissions (0600), which is a standard and secure practice for command-line tools.
[Subprocess Execution]: The skill utilizes Python's subprocess module in scripts/archive.py to interface with the local git CLI for respecting .gitignore rules, and in scripts/auth.py to manage a local authentication callback server. The arguments passed to these subprocesses are controlled by the skill's logic, mitigating common injection risks.
[Remote Resource Validation]: When fetching application code from GitHub, the scripts/archive.py script performs extensive validation of the downloaded ZIP archive. This includes checking for size limits, entry counts, compression ratios, and the absence of symbolic links to protect against common archive-based security considerations.
[Verified Infrastructure]: The skill interacts with backend services hosted on official AWS domains (*.api.aws and *.amazonaws.com). The scripts/launch_config.py script includes validation logic to ensure that configuration overrides for these endpoints must still target the official AWS domain ecosystem.

launch-with-aws