The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes application source code and manifest files (e.g., package.json, requirements.txt) to detect stacks and frameworks. This data influences the skill's choice of integration documentation and code generation. While necessary for its function, this creates a surface where malicious content in the analyzed code could potentially influence agent actions. The skill lacks explicit boundary markers for this untrusted data but requires human confirmation for file mutations.
[REMOTE_CODE_EXECUTION]: The skill uses uvx to run arize-tracing-assistant@latest. This involves downloading and executing code from a remote source. This package is maintained by the skill's author, Arize AI, and is part of the intended developer experience.
[COMMAND_EXECUTION]: The skill executes ax CLI commands for profile management and trace inspection. These commands are used to interact with the local environment and Arize's backend services in a controlled manner.
[EXTERNAL_DOWNLOADS]: The skill downloads documentation and configuration from arize.com and installs instrumentation libraries from standard package registries. These resources are provided by the vendor and well-known open-source organizations.
[PERSISTENCE]: The skill may persist configuration by updating ax profiles or modifying shell startup files like .zshrc. These actions are performative and require explicit user consent as per the skill's instructions.

arize-instrumentation