The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the @headlessx-cli/core package via npm or pnpm from the public registry. This is the primary method for deploying the tool's capabilities.
[COMMAND_EXECUTION]: The skill makes extensive use of local command execution. It uses subprocess.run in scripts/smoke_cli.py to verify binary availability and help outputs. Furthermore, it utilizes the headlessx CLI for critical operations such as workspace initialization (init), service management (start, stop), and logging (logs).
[INDIRECT_PROMPT_INJECTION]: The skill possesses a significant attack surface for indirect prompt injection because it is designed to ingest untrusted data from the open web.
Ingestion points: Functions like headlessx scrape, headlessx map, headlessx crawl, and headlessx google fetch content from external URLs and search results into the agent's context (referenced in SKILL.md and references/command-matrix.md).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands were found in the provided documentation for data ingestion.
Capability inventory: The skill has high-privilege local capabilities, including file system writes (-o page.md), network operations, and Docker management (headlessx restart for production modes).
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from websites before it is presented to the agent.

cli