The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references several external resources including official documentation (docs.wandb.ai), GitHub repositories (github.com/wandb/wandb), and PyPI. All domains are official and trusted for the 'wandb' ecosystem. Additionally, it references well-known technology vendors such as OpenAI, Anthropic, and Hugging Face in the context of integrations.
[COMMAND_EXECUTION]: Instructions include standard package installation commands (pip install wandb) and CLI usage for authentication (wandb login). These are expected for a development tool and follow security best practices by advising the use of environment variables for API keys.
[DATA_EXPOSURE]: The documentation provides clear warnings against committing API keys to source control and correctly identifies that the wandb login command stores credentials in the standard ~/.netrc location.
[DYNAMIC_EXECUTION]: The skill contains a Python script scripts/check-updates.py. Analysis of the source code confirms it is a legitimate maintenance utility that queries the PyPI JSON API to check for the latest package version and verifies local file integrity. It does not perform any unauthorized data exfiltration or arbitrary command execution.
[PROMPT_INJECTION]: The SKILL.md file uses mandatory triggers and instructions to ensure the agent uses this skill for relevant ML/LLM tasks. These are standard instructional parameters and do not attempt to bypass agent safety filters or override core system instructions.

wandb