The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructions facilitate processing data from external sources which could contain malicious instructions.
Ingestion points: Data ingestion occurs through requests.get from external APIs (references/examples.md) and reading files from cloud storage like S3 (references/examples.md).
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the provided examples.
Capability inventory: The skill possesses extensive capabilities including cloud code execution via modal run and modal deploy, system command execution in containers via run_commands (references/images.md), and persistent storage operations via Modal Volumes.
Sanitization: The examples do not demonstrate sanitization or validation of the ingested external data before processing.
[COMMAND_EXECUTION]: The skill uses the Modal CLI for managing cloud resources.
Evidence: Usage of modal run, modal deploy, and modal token new is central to the skill's functionality (SKILL.md, references/getting-started.md).
[EXTERNAL_DOWNLOADS]: The skill configuration and examples involve downloading packages and assets from external repositories.
Evidence: Fetches Python packages via uv pip install and modal.Image.uv_pip_install. Clones repositories from GitHub (e.g., github.com/modal-labs/agi) during image construction (references/images.md).

modal