The Agent Skills Directory

[COMMAND_EXECUTION]: The skill employs dynamic context injection in the SKILL.md file to retrieve the current git status and directory structure. These commands (git branch, git status, and find) are benign, limited in scope, and used solely to provide necessary context for the agent to operate within the local environment.
[PROMPT_INJECTION]: The skill functions by ingesting external data from URLs and GitHub repositories, which represents a potential surface for indirect prompt injection. However, this is the primary intended purpose of the skill and is triggered by explicit user input.
Ingestion points: Step 2 in SKILL.md specifies reading from external URLs, GitHub repositories (via gh), and local sibling paths provided in the $ARGUMENTS.
Boundary markers: No explicit delimiters or instruction-isolation warnings are present for the external data being studied.
Capability inventory: The skill uses gh for repository interaction, web fetching for external docs, and file system access for reading and writing code to the local repository.
Sanitization: The instructions do not specify validation or sanitization of the fetched reference material before it is processed by the agent.

mirror-reference