The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses subprocess to execute the dtctl CLI tool for legitimate administrative tasks such as querying settings, fetching document bodies, and applying new configurations. The command arguments are strictly controlled by the script logic and do not allow for arbitrary shell injection.
[INDIRECT_PROMPT_INJECTION]: The skill recognizes the risk of indirect prompt injection from untrusted data in the Dynatrace environment (e.g., service or user names). It implements a sanitization layer in scripts like actors.py that filters control characters and enforces length limits on external strings before they are processed by the agent.
[CREDENTIALS_UNSAFE]: The skill handles sensitive platform and classic API tokens. It incorporates redaction logic in its configuration and error handling modules (e.g., using a SecretToken class and masking functions) to ensure tokens are never logged or exposed in plain text within session transcripts.
[DYNAMIC_EXECUTION]: The skill includes several Python and TypeScript/JavaScript scripts for complex migration tasks. These tools are focused strictly on configuration translation and tenant auditing, and they do not involve arbitrary code generation or execution from untrusted sources.
[EXTERNAL_DOWNLOADS]: The skill refers to downloading the dtctl utility and extension artifacts from trusted vendor domains (dynatrace.com) and official repositories (GitHub). These downloads are conducted securely as part of the intended administrative functionality.

dt-upgrade