The Agent Skills Directory

[SAFE]: The skill correctly implements Datadog LLM Observability using official SDKs (ddtrace for Python and dd-trace for Node.js). All configuration follows official Datadog documentation for agentless and agent-based tracing.
[COMMAND_EXECUTION]: The skill utilizes local shell commands (such as python -c, node -e, and ddtrace-run) specifically for verifying the successful installation and import of the instrumentation packages. These are standard development practices for confirming a library is correctly configured in the runtime environment.
[CREDENTIALS_UNSAFE]: The skill manages the DD_API_KEY by utilizing a local tool (CreateApiKey) and persisting the value to standard local environment files (.env, .env.local). This is appropriate for the skill's purpose of setting up a development environment and does not involve hardcoded secrets or insecure exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill contains logic to extract session identifiers from incoming web request bodies or headers. While this is an ingestion of untrusted data, it is used strictly for telemetry tagging (session tracking) and includes secure fallbacks to system-generated UUIDs, which is consistent with secure observability patterns.

dd-instrument-llmo