The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill utilizes npx @arizeai/phoenix-cli to execute the Phoenix CLI, which involves fetching the package from the npm registry if not already present. It also includes a px docs fetch command that downloads documentation from the configured Phoenix host to the local filesystem.
[COMMAND_EXECUTION]: The skill defines various shell commands for interacting with the Phoenix service, including trace management, GraphQL API queries, and documentation fetching. These commands are standard for the intended debugging and observability use case.
[CREDENTIALS_UNSAFE]: The documentation mentions PHOENIX_API_KEY=your-api-key as a setup step. This is a placeholder for environment variable configuration, which is a recommended security practice for secret management, and no actual credentials are exposed.
[PROMPT_INJECTION]: The skill includes functionality to fetch prompt templates (px prompt get) and documentation (px docs fetch) from an external Phoenix server. While this introduces a surface for indirect prompt injection if the source data is compromised, the skill does not contain any direct malicious instructions.

phoenix-cli