The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides instructions for processing data from the Box API, such as file metadata, listings, and content, which introduces a surface for indirect prompt injection if that external data contains malicious instructions.
Ingestion points: Data is retrieved through SDK calls (e.g., legacy Client.file or generated client methods) as described in SKILL.md and references/file-transfer.md.
Boundary markers: The skill does not explicitly require the agent to use delimiters or ignore embedded instructions when interpreting these inputs.
Capability inventory: The skill facilitates network operations for API interaction and file system writes for local downloads, as referenced in references/file-transfer.md and references/python.md.
Sanitization: Guidance includes verifying file integrity using SHA-1 and byte counts, as well as omitting credentials and sensitive content from logs.
[EXTERNAL_DOWNLOADS]: The skill references official Box SDKs and documentation for installation and development purposes.
Fetches resources and documentation from official sources, including the box-node-sdk and box-python-sdk repositories under the box GitHub organization.
Refers to standard installation procedures via package managers like npm for official SDK distributions.

box-sdk-development