The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the repomix command to summarize GitHub repositories. It also specifies writing files to the .claude/skills/$1 directory based on user-provided arguments.
[EXTERNAL_DOWNLOADS]: The skill provides an external link to repomix.com and instructs the agent to install the tool if it is not already present on the system.
[PROMPT_INJECTION]: The skill uses strong instructional markers such as [IMPORTANT], MUST FOLLOW, and Your mission to control agent behavior. While these are common in skill definitions, they are noted as behavioral overrides.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill ingests untrusted data from a user-provided prompt ($2), external web URLs, and content from GitHub repositories.
Boundary markers: The skill uses XML-like tags (<reference-or-script-prompt>) to wrap the secondary argument, providing some structural separation.
Capability inventory: The skill has the capability to write files to the local file system (.claude/skills/), execute shell commands (repomix), and spawn subagents to explore external content.
Sanitization: No explicit sanitization, validation, or escaping of the ingested external content is mentioned before it is processed or used to generate new files.

skill-add