The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes external project files to determine the necessary instrumentation logic.
Ingestion points: Reads dependency manifests (package.json, pyproject.toml, requirements.txt, setup.py, Pipfile, pom.xml, build.gradle, go.mod) and source code imports to detect frameworks and providers.
Boundary markers: The skill does not explicitly define markers to separate manifest content from its internal instructions.
Capability inventory: The skill has the capability to install packages, execute shell commands via the ax CLI, and write instrumentation code to the local filesystem.
Sanitization: No explicit validation or filtering of external manifest content is described before it is used to drive implementation decisions.
[COMMAND_EXECUTION]: The skill invokes system commands to configure the environment and manage software dependencies.
Executes ax CLI commands for profile management, specifically ax profiles show, ax profiles create, and ax profiles update as described in references/ax-profiles.md.
Uses package managers (pip, npm, go get) to install instrumentation libraries and exporters.
[EXTERNAL_DOWNLOADS]: The skill fetches resources from external domains during the instrumentation process.
Downloads technical documentation and integration guides from the arize.com domain.
Installs Arize-specific and OpenTelemetry packages from public package registries like npm and PyPI.
[PERSISTENCE]: The skill recommends persisting configuration variables in shell environment files to maintain settings across sessions.
Instructs the user to add export ARIZE_SPACE and export ARIZE_API_KEY to shell startup scripts such as ~/.zshrc or ~/.bashrc in the references/ax-profiles.md file.

arize-instrumentation