The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill guides the agent to process repository files like justfiles and project configurations to automate development tasks. This creates a surface where an attacker who can modify repository files could potentially inject instructions into the agent's workflow.
Ingestion points: The agent is instructed to read existing justfiles, check command outputs (just --list), and evaluate repository structure within the workspace.
Capability inventory: The skill enables the agent to write and execute just recipes, which can include shell commands, inline scripts (Bash, Python, Node.js), and network operations via tools like curl or aws CLI.
Sanitization: The instructions emphasize manual inspection of expanded commands and the use of dry-run modes to ensure safety before execution.
Boundary markers: The skill does not prescribe specific delimiters or safety prompts for the agent to use when processing potentially untrusted content from the repository.
[COMMAND_EXECUTION]: The skill facilitates the execution of shell commands and scripts through the just task runner. While this is the intended functionality for task automation, it provides the agent with the ability to perform significant operations in the user's environment, including file modifications and tool invocations.

cli-just