The Agent Skills Directory

[COMMAND_EXECUTION]: The skill attempts to locate and execute a binary named browse by searching several hidden paths, including .claude/skills/lazyweb-skill/browse/dist/browse and ~/.claude/skills/gstack/browse/dist/browse. It also executes git rev-parse --show-toplevel to determine repository boundaries.\n- [CREDENTIALS_UNSAFE]: The skill reads ~/.lazyweb/libraries.json, which is a configuration file likely containing sensitive search session information or authentication details for connected inspiration libraries.\n- [DYNAMIC_EXECUTION]: The specific path to the browse tool is determined at runtime via shell script logic, and the resulting variable is then executed with various subcommands and parameters.\n- [EXTERNAL_DOWNLOADS]: The skill uses curl to download image files from remote URLs retrieved from the Lazyweb search API.\n- [INDIRECT_PROMPT_INJECTION]:\n
Ingestion points: Processes content from external websites via the browse tool and consumes visionDescription metadata from Lazyweb search results.\n
Boundary markers: No specific delimiters or boundary instructions are used to separate external data from agent instructions.\n
Capability inventory: Includes shell command execution, file system read/write, and network search capabilities.\n
Sanitization: No explicit sanitization or filtering is performed on external metadata or website content before it is interpolated into the generated markdown and HTML reports.

lazyweb-quick-references