The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation in assets/parallel-universes.mdx includes instructions to download and install the ollama Python package via pip for local inference.
[COMMAND_EXECUTION]: The skill provides Python code examples in assets/parallel-universes.mdx for interacting with S3-compatible object storage via the boto3 library. The examples demonstrate creating buckets, managing snapshots, and forking datasets using environment variables for credential management, which is consistent with standard development practices.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest untrusted user data in the form of "brain dumps" or outlines to generate blog posts.
Ingestion points: User-provided notes or outlines accepted as the primary input for the style transformation task described in SKILL.md.
Boundary markers: The instructions do not specify unique delimiters or specific instructions to ignore embedded commands within the user's provided notes.
Capability inventory: The skill's instructions are limited to text generation and prose transformation; no automated scripts are provided that execute system commands or perform network operations at runtime based on the untrusted data.
Sanitization: The skill lacks explicit sanitization or filtering logic for instructions embedded within the user's brain dump, relying instead on the model's core safety guardrails.

xe-writing-style