The Agent Skills Directory

[DATA_EXPOSURE_AND_EXFILTRATION]: The skill manages sensitive Datadog credentials (API and Application keys) with several built-in security controls. It reads from and writes to .env or .env.local files, enforcing strict 0600 file permissions and ensuring these files are added to .gitignore to prevent accidental leakage. Secrets are consistently masked in all logs and user-facing outputs, displaying only the last four characters.
[COMMAND_EXECUTION]: The skill utilizes a one-shot Python script to handle local OAuth callbacks. This script is embedded as a static heredoc, using standard library modules (http.server) to catch the authentication code locally, which is a secure and standard practice for CLI-based authentication. Additionally, the skill uses eval for variable resolution in shell loops, but only against a hardcoded list of configuration keys (DD_SITE, DD_API_KEY, DD_APP_KEY), preventing arbitrary command injection.
[EXTERNAL_DOWNLOADS]: The skill performs network requests to ipinfo.io to facilitate automatic region detection based on the user's IP address. This is used solely for user convenience during the onboarding process and does not involve the transmission of sensitive data.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external sources, including geolocation responses and OAuth callback parameters. It mitigates potential injection risks by using specific parsing utilities (grep/sed) and providing an escaping function for JSON construction to ensure data is handled as literal strings rather than executable instructions.

dd-account-setup