dd-orchestrator
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches configuration and task-specific instructions from Datadog's official GitHub repository (
datadog-labs/agent-skills) and Datadog's onboarding API. These downloads are restricted to verified Datadog infrastructure to ensure the agent uses the most current and authoritative setup logic. - [DATA_EXFILTRATION]: Transmits operational telemetry, such as session status and step outcomes, to Datadog's monitoring backend. The reported data is restricted to an allow-list of non-sensitive metadata, including environment types and organization public identifiers, used for tracking onboarding success.
- [COMMAND_EXECUTION]: Uses local Python utilities and shell commands to analyze the repository context, resolve the ordering of setup tasks, and communicate with Datadog APIs. These operations are scoped to standard project detection and telemetry reporting.
- [INDIRECT_PROMPT_INJECTION]: Analyzes user-provided goals and project files to recommend monitoring products. The skill mitigates the risk of malicious instructions in project data by requiring the user to manually review and approve the generated plan before any skills are dispatched.
- [DYNAMIC_EXECUTION]: Downloads and executes specialized task instructions (skills) at runtime. This dynamic loading is restricted to verified sources owned by the vendor, providing a modular and up-to-date onboarding experience.
Audit Metadata