The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The scripts/agent.py script accepts Datadog API and Application keys as command-line arguments (sys.argv[1] and sys.argv[2]). This practice can expose sensitive secrets in process listings (e.g., ps aux) and shell history.
[DATA_EXFILTRATION]: The skill collects high-value reconnaissance data including security detection rules, monitor states, and security signal summaries. This aggregated metadata is written to a local file (datadog_security_report.json) and printed to stdout, which could be exploited if the agent environment is compromised.
[COMMAND_EXECUTION]: The skill executes local Python scripts to interact with the Datadog API and generate security reports.
[EXTERNAL_DOWNLOADS]: The skill references documentation and source code from Datadog and GitHub, which are recognized well-known services for security operations.
[PROMPT_INJECTION]: Indirect Prompt Injection Surface (Category 8): * Ingestion points: API responses from security monitoring and monitor endpoints in scripts/agent.py. * Boundary markers: Absent; findings are printed directly to the console. * Capability inventory: Network requests via the requests library and file system write access. * Sanitization: The agent does not sanitize or escape strings retrieved from the API (such as monitor names or signal details) before including them in its output, potentially allowing malicious content from the API source to influence downstream agent processing.

implementing-security-monitoring-with-datadog