The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill reads from and writes to sensitive local files (.env, .env.local) to manage DD_API_KEY and DD_APP_KEY. It mitigates risk by setting file permissions to 0600 and using umask 077, while ensuring credentials are masked in user-facing output.
[EXTERNAL_DOWNLOADS]: Makes network requests to ipinfo.io to automatically suggest the appropriate Datadog region based on the user's location.
[COMMAND_EXECUTION]: Utilizes bash and curl to execute authentication flows, perform API validation, and update local environment configurations.
[DATA_EXFILTRATION]: Transmits session tokens and identity information to Datadog-owned domains (datadoghq.com, datadoghq.eu, etc.) as part of the standard OAuth and account creation process.
[DYNAMIC_EXECUTION]: Assembles and executes an inline Python script at runtime to create a temporary local HTTP server for capturing OAuth redirect callbacks.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external APIs which could theoretically contain malicious instructions. Ingestion points: Geolocation data from ipinfo.io and identity data from Datadog API responses (e.g., /api/v2/current_user) in SKILL.md and references/get-api-key.md. Boundary markers: The flow relies on user confirmation via the host's native selector before committing to detected regions. Capability inventory: bash and curl networking (all scripts), sensitive file-write (SKILL.md, references/get-api-key.md), and python3 script execution (references/authenticate.md). Sanitization: Implements an esc() function to escape special characters in user-provided fields before inclusion in JSON payloads.

dd-account-setup