The Agent Skills Directory

[DATA_EXFILTRATION]: The skill follows secure credential handling practices by requiring the use of environment variables ($ARIZE_API_KEY) for authentication and explicitly prohibiting the agent from requesting, displaying, or logging raw secrets in the conversation context.
[COMMAND_EXECUTION]: Administrative tasks are performed via the ax CLI. The skill enforces a 'Destructive-action rule' that requires the agent to explain the impact and obtain explicit user confirmation via the AskUserQuestion tool before executing any deletion, revocation, or removal commands with the --force flag.
[EXTERNAL_DOWNLOADS]: Setup instructions refer to the installation and upgrade of the vendor's official CLI package (arize-ax-cli) from standard package registries using tools like pip, pipx, or uv.
[INDIRECT_PROMPT_INJECTION]: The skill processes structured JSON data from CLI outputs (Ingestion points: 'ax ... list -o json' in SKILL.md). While explicit boundary markers and sanitization are absent, the security risk is mitigated by the restricted operational scope and the mandatory human-in-the-loop confirmation for all administrative changes (Capability inventory: resource deletion, user invitation, and role management across all reference scripts).

arize-admin