The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by design. It is instructed to read and report on external data sources that are not fully controlled by the agent.
Ingestion points: Processes content from application logs (storage/logs/), browser console logs via agent-browser, and database records.
Boundary markers: The instructions lack specific delimiters or instructions to ignore potential commands embedded within the logs or database content being analyzed.
Capability inventory: The agent has the ability to execute artisan console commands, run npm build scripts, and interact with the filesystem and browser.
Sanitization: There is no evidence of filtering or sanitizing the data ingested from logs before it is processed by the agent's logic.
[COMMAND_EXECUTION]: The skill is explicitly designed to execute shell commands within the project environment.
Evidence: Instructions include running artisan commands with realistic arguments and executing npm run build to address Vite manifest errors. While these are standard developer workflows, they represent a significant capability tier that could be abused if the agent is influenced by malicious input.

verify