The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the npx package runner, specifically for installing tools with flags that bypass confirmation (npx impeccable install -y --scope=project --providers=claude).
[REMOTE_CODE_EXECUTION]: The instructions require downloading and installing the impeccable package from an external source at runtime. This involves fetching and executing third-party code that has not been locally verified or pinned to a specific version.
[DYNAMIC_EXECUTION]: The skill uses a tool called hallmark to perform a "redesign" of the project. This process involves the dynamic generation of a design.md configuration and the automated rewriting of project source code files based on newly generated design tokens.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to audit and process untrusted project data, such as route configurations and existing UI source code, which could contain hidden malicious instructions targeted at the agent.
Ingestion points: Project source code files, router/pages configuration files, and live UI views captured via Playwright.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded prompts within the project files.
Capability inventory: The agent has access to shell execution (npx), file system modifications (hallmark redesign), and network operations (playwright-verify and /qc-uiux).
Sanitization: There are no specified sanitization steps for the data ingested from the project files before it is processed by the design and audit engines.

design-prim