The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the claudish package globally via npm install -g claudish. This is a standard dependency for the skill's visual analysis functionality.
[COMMAND_EXECUTION]: The skill makes extensive use of shell commands to execute the claudish CLI tool for processing screenshots through external vision models. It also utilizes the javascript_tool to execute arbitrary scripts within the browser context to facilitate debugging.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks as it processes untrusted data from external websites.
Ingestion points: Untrusted data enters the agent context through mcp__claude-in-chrome__read_page, mcp__claude-in-chrome__read_console_messages, and mcp__claude-in-chrome__read_network_requests tools used in various recipes.
Boundary markers: No specific boundary markers or instructions to ignore embedded commands are present in the recipes provided.
Capability inventory: The agent has access to javascript_tool for browser-side execution and shell execution via claudish as well as network access through browser navigation tools.
Sanitization: No sanitization or filtering of the ingested browser content is performed before it is presented to the agent or external vision models.

browser-debugging