The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external repository files, such as YAML configurations and Fluid templates, which could be controlled by an attacker. Malicious instructions placed in these files could potentially influence the agent's actions during the audit or overhaul process.
Ingestion points: The scripts/audit-content-elements.php script reads and parses config.yaml, fixture.json, and frontend.fluid.html files from the target repository.
Boundary markers: Absent. The files are read and their contents processed without specific delimiters to segregate untrusted data from instructions.
Capability inventory: The skill instructs the agent to execute shell commands including npx, php, and ripgrep (rg).
Sanitization: Absent. The script and search commands perform analysis on raw content without specific security filtering for LLM context.
[COMMAND_EXECUTION]: The skill instructions require the agent to execute various shell commands. These include project-specific PHP audit scripts (php scripts/audit-content-elements.php), search utilities (rg), and the shadcn CLI (npx shadcn). While these are standard tools for the described development workflow, they represent a powerful capability accessible via potentially untrusted input paths.
[EXTERNAL_DOWNLOADS]: The skill utilizes npx shadcn@latest to fetch and run the shadcn/ui CLI tool. This is a well-known service used for managing UI components and design system tokens in modern web development, and the skill uses it to build local registries and retrieve preset configurations.

typo3-shadcn-content-elements