The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use the Bash tool for project maintenance and development tasks. This includes inspecting directory structures (e.g., ls src/components/ds/atoms/) and running project-specific scripts like bun run update-icons to regenerate type definitions after icon updates.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for processing user-provided data (SVG content). Specifically, it instructs the agent to use a shell pipe (echo '<svg>...</svg>' | bunx svgo ...) to optimize the SVG. This creates an attack surface where maliciously crafted SVG content containing shell metacharacters could attempt to escape the command string and execute arbitrary commands. While the instructions suggest using single quotes for wrapping the input, this remains a potential vector for command injection.
Ingestion points: User-pasted SVG content in the chat interface.
Boundary markers: The instructions suggest wrapping user content in single quotes within the echo command.
Capability inventory: The skill has access to the Bash tool and can execute shell commands, including those involving pipes and external package runners (bunx).
Sanitization: There is no explicit instruction to sanitize or escape shell-sensitive characters (like backticks or single quotes) within the user-provided SVG content before execution.
[EXTERNAL_DOWNLOADS]: The skill uses bunx svgo to perform SVG optimization. svgo is a widely recognized and well-known utility for optimizing SVG files, and its execution via a package runner is a standard practice in modern web development workflows.

openbb-design