The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches research paper content and AI-generated overviews from the AlphaXiv service (alphaxiv.org) using curl commands as defined in references/fetching-papers.md.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external research papers which is then used to guide the generation of executable notebooks and widgets.
Ingestion points: External paper content is retrieved via curl from alphaxiv.org in references/fetching-papers.md.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" markers to prevent the agent from being influenced by malicious text embedded within the research papers.
Capability inventory: The agent is authorized to generate marimo notebooks (Python), anywidget components (JavaScript), and execute network commands (curl).
Sanitization: There is no explicit sanitization or validation of the fetched markdown content before it is interpolated into the agent's context for code generation.
[DYNAMIC_EXECUTION]: The skill generates JavaScript for anywidget components and Python code for marimo notebooks based on external input. While the instructions in references/anywidget.md provide a safety baseline by prohibiting access to files outside the project (such as ~/.ssh or .env), the generation of executable scripts from untrusted data is an inherent risk vector.

implement-paper