The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses various shell commands to facilitate its workflow, such as curl to verify server connectivity and standard Unix utilities like find, wc, and grep to inventory and process audit reports. Additionally, it uses git status to verify the output directory's state. These commands are used for internal coordination and inventory purposes within the intended scope of the skill.
[INDIRECT_PROMPT_INJECTION]: The skill facilitates the processing of data from external web applications through browser-based sub-agents. This creates a potential surface where untrusted content from audited sites could attempt to influence the agent's reasoning or reporting.
Ingestion points: Audited web application content accessed via the run-agent-browser tool (SKILL.md, references/persona-subagent-template.md).
Boundary markers: No explicit delimiters or safety instructions are defined to separate ingested web content from the agent's instructional context.
Capability inventory: The skill is capable of writing markdown findings and screenshots to the local file system and dispatching sub-agents using the platform's Agent tool.
Sanitization: There are no specific sanitization or filtering procedures described for the content retrieved from audited web pages.
[EXTERNAL_DOWNLOADS]: The skill's documentation provides instructions for installing the skill and its parent pack from the author's GitHub repository using the npx package runner. This targets the vendor's official resources and follows standard installation practices for the platform.

audit-ux-and-save-files