The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references the official Julia language website and documentation (julialang.org), as well as reputable ecosystem project sites like sciml.ai, juliastats.org, and jump.dev. These references are standard for developer documentation and do not point to untrusted or malicious remote sources.
[COMMAND_EXECUTION]: The reference file EXAMPLES.md provides common shell commands for Julia installation, package management (using the ']' package mode), and script execution. These are benign instructional examples intended for user environment setup.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of user-supplied Julia code and scientific queries, creating a standard surface for indirect prompt injection. 1. Ingestion points: The skill analyzes Julia code provided in user prompts or referenced files. 2. Boundary markers: None; there are no specific instructions to the agent to treat user-provided code blocks as untrusted data or use delimiters. 3. Capability inventory: The skill is permitted to use the Bash, Read, Write, Edit, Glob, and Grep tools as defined in the frontmatter. 4. Sanitization: None; the instructions do not include logic for filtering or escaping input data.

julia-expert