The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions and examples for the agent to execute shell commands via the just task runner. It facilitates common development operations such as building, testing, and cleaning directories through recipes defined in project files.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow where the agent ingests and executes content from external justfile definitions.
Ingestion points: Instructions in SKILL.md direct the agent to read and list recipes from justfile documents located in the project root or specified subdirectories.
Boundary markers: The instructions do not specify boundary markers or prompt delimiters to mitigate the risk of the agent interpreting recipe content or output as new instructions.
Capability inventory: The skill enables the execution of various shell commands including testing tools (pytest), package scripts (npm run), linter checks (ruff), and system file operations (rm) as seen in the example recipes in SKILL.md.
Sanitization: No validation or sanitization mechanisms are implemented for the commands defined in the external task files.

task-runner