The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The scripts/install.sh script downloads platform-specific Lightpanda binaries from the author's official GitHub repository (lightpanda-io/browser). The installer implements a security check by fetching and verifying the SHA256 checksum of the binary via the GitHub API before granting execution permissions.
[COMMAND_EXECUTION]: The skill executes the lightpanda binary to support its primary functions, including acting as an MCP (Model Context Protocol) server, performing CLI-based web fetches, and running a CDP (Chrome DevTools Protocol) server for automation.
[REMOTE_CODE_EXECUTION]: The skill provides an evaluate tool and custom CDP methods that allow the agent to execute JavaScript code within the browser's page context. This is expected functionality for a web automation and scraping tool.
[INDIRECT_PROMPT_INJECTION]: The skill provides an interface to ingest and process content from arbitrary external URLs, creating a surface for indirect instructions to reach the agent.
Ingestion points: Tools such as goto, markdown, and fetch (documented in SKILL.md) ingest raw and processed data from web pages.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the documentation or tool descriptions.
Capability inventory: The skill enables JavaScript execution (evaluate), form interaction (fill), and element clicking (click).
Sanitization: The skill returns page content (HTML/Markdown) to the agent without explicit sanitization or filtering of potential injection patterns.

Lightpanda